CA2083074A1 - Defect detection - Google Patents
Defect detectionInfo
- Publication number
- CA2083074A1 CA2083074A1 CA002083074A CA2083074A CA2083074A1 CA 2083074 A1 CA2083074 A1 CA 2083074A1 CA 002083074 A CA002083074 A CA 002083074A CA 2083074 A CA2083074 A CA 2083074A CA 2083074 A1 CA2083074 A1 CA 2083074A1
- Authority
- CA
- Canada
- Prior art keywords
- defect
- pixel values
- detection system
- pixel
- defect detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Abandoned
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/0006—Industrial image inspection using a design-rule based approach
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/89—Investigating the presence of flaws or contamination in moving material, e.g. running paper or textiles
- G01N21/8901—Optical details; Scanning details
- G01N21/8903—Optical details; Scanning details using a multiple detector array
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
- G06T2207/10021—Stereoscopic video; Stereoscopic image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30124—Fabrics; Textile; Paper
Abstract
Abstract of the Disclosure A defect detection system includes a video camera with defect detection circuits for detecting defects in video signals being outputted by corresponding sections of an array sensor such as a TDI CCD two-dimensional array sensor. Each defect detection circuit includes a subtraction circuit for subtracting a prior stored pixel from an incoming pixel to generate a difference.
Comparators compare the difference with positive and negative limits defining an acceptable range of difference values. The prior stored pixel is updated to the succeeding pixel only when the difference value is acceptable. Memories store the defect pixels from the respective detection circuits along with X-coordinates and end of line bits. The memories are sequentially read up to their end of line bits, and the defect pixel values along with coordinates expanded to include section indicating bits are transferred from the camera to further processing facilities.
Comparators compare the difference with positive and negative limits defining an acceptable range of difference values. The prior stored pixel is updated to the succeeding pixel only when the difference value is acceptable. Memories store the defect pixels from the respective detection circuits along with X-coordinates and end of line bits. The memories are sequentially read up to their end of line bits, and the defect pixel values along with coordinates expanded to include section indicating bits are transferred from the camera to further processing facilities.
Description
`~:
`~ 104-1031~ 2~3~7~
`:. DWM
.~ DEFECT DETECTION
. .
-` Technical Field ~` The present invention relates to defect detection ;C using video cameras, for example, the detection of a '~ 5 defect in a continuous web using a time delay and ~` integration (TDI) charge ~oupled device (CCD).
Back~round rt ~- The prior art discloses several de~ect detection '~ devices and processes using video cameras to produce a :-~
video signal which is then analyzed to detect a defect in the objects viewed by the cameras. International Application Publication No. WO 91/14173 published ~` September 19, 1991, discloses the use of a plurality of cameras, each containing a TDI CCD array sensor, arranged in a line across a moving web for generating a plurality of continuous video signals representing images o~ respective portions o* the moving web. These ` video signals are transmitted from the cameras to respective processing units which analyze the vidQo signals for defects. The analog outputs of the TDI CCD
array sensors are converted to streams of binary digital '~ pixel signals either in the camera unit or in the processing units. The digital pixel signals are `~$ temporarily stored in computer memories of the ,~ 25 processing units until they are analyzed by computers in the processing units to de~ermine the presence o~ a ~; defect.
~ One algorithm employed by the computers in the `~ processing units of the above mentioned International Application No. WO 91/14173 sequentially compares each digital pixel value of a line image across a web portion with a dynamically adjustable range. I* a present pixel value is outside of the dynamic range but within ;~ predetermined ranges above or below the previously tested prior acceptable pixel value , ~`
.
:
, ,, .
, . . - .
`~ 104-1031~ 2~3~7~
`:. DWM
.~ DEFECT DETECTION
. .
-` Technical Field ~` The present invention relates to defect detection ;C using video cameras, for example, the detection of a '~ 5 defect in a continuous web using a time delay and ~` integration (TDI) charge ~oupled device (CCD).
Back~round rt ~- The prior art discloses several de~ect detection '~ devices and processes using video cameras to produce a :-~
video signal which is then analyzed to detect a defect in the objects viewed by the cameras. International Application Publication No. WO 91/14173 published ~` September 19, 1991, discloses the use of a plurality of cameras, each containing a TDI CCD array sensor, arranged in a line across a moving web for generating a plurality of continuous video signals representing images o~ respective portions o* the moving web. These ` video signals are transmitted from the cameras to respective processing units which analyze the vidQo signals for defects. The analog outputs of the TDI CCD
array sensors are converted to streams of binary digital '~ pixel signals either in the camera unit or in the processing units. The digital pixel signals are `~$ temporarily stored in computer memories of the ,~ 25 processing units until they are analyzed by computers in the processing units to de~ermine the presence o~ a ~; defect.
~ One algorithm employed by the computers in the `~ processing units of the above mentioned International Application No. WO 91/14173 sequentially compares each digital pixel value of a line image across a web portion with a dynamically adjustable range. I* a present pixel value is outside of the dynamic range but within ;~ predetermined ranges above or below the previously tested prior acceptable pixel value , ~`
.
:
, ,, .
, . . - .
- 2 ~
then the upper and lower limi~s o~ ~h~ ~ynamic range are changed up or down, respectively, by the difference between the present pixel value and the prior pixel value. This compensates for variations in thickness o~
the web, gradual increases and decreases in pixel values, which are normal and differentiate such normal changes from defects which create sudden changes greater than the predetermined ranges.
One deficiency of prior art defect detecting systems employing video cameras concerns the transmission of analog or digital signals from the cameras to the processing units and the temporary storage of these signals in the processing units. In web defect detection at relatively high resolution, required transmission rates can be up to or higher than thirty-two million bytes or pixels per second per camera. Reliable transmission and handling of a plurality of video signals at such frequencies is difficult and requires relatively expensive facilities.
Large (several megabyte) high speed dual ported video memories with facilities to properly address and store incoming signals are required. These memories and their high speed computer processors are relatively complex and expensive. Even with such expensive high speed equipment, there are often unacceptable limitations on inspection rates and/or defect detection resolution.
Summary o~ Invention In a first aspect, the invention is summarized in a defect detection system having a defect detecting circuit mounted in a camera housing along with an array sensor. An image of an object under test is projected by a lens onto the array sensor. The defect detecting circuit detects pixels of an object defect in a video signal generated by the array sensor. A camera output transmits the detected defect pixels whereby the quantity of pixel values transmitted by the camera output for further processing is less than the quantity .~ ~
then the upper and lower limi~s o~ ~h~ ~ynamic range are changed up or down, respectively, by the difference between the present pixel value and the prior pixel value. This compensates for variations in thickness o~
the web, gradual increases and decreases in pixel values, which are normal and differentiate such normal changes from defects which create sudden changes greater than the predetermined ranges.
One deficiency of prior art defect detecting systems employing video cameras concerns the transmission of analog or digital signals from the cameras to the processing units and the temporary storage of these signals in the processing units. In web defect detection at relatively high resolution, required transmission rates can be up to or higher than thirty-two million bytes or pixels per second per camera. Reliable transmission and handling of a plurality of video signals at such frequencies is difficult and requires relatively expensive facilities.
Large (several megabyte) high speed dual ported video memories with facilities to properly address and store incoming signals are required. These memories and their high speed computer processors are relatively complex and expensive. Even with such expensive high speed equipment, there are often unacceptable limitations on inspection rates and/or defect detection resolution.
Summary o~ Invention In a first aspect, the invention is summarized in a defect detection system having a defect detecting circuit mounted in a camera housing along with an array sensor. An image of an object under test is projected by a lens onto the array sensor. The defect detecting circuit detects pixels of an object defect in a video signal generated by the array sensor. A camera output transmits the detected defect pixels whereby the quantity of pixel values transmitted by the camera output for further processing is less than the quantity .~ ~
3 2~,3~
of pixel values in the video signal, for example, by a ratio of 1:1000 or less of defect pixels to total pixels.
In a second aspect, the invention is summarized in a defect detection system having a subtraction circuit along with a comparison circuit. A prior pixel value derived from one or more pixels prior to a present pixel value in a video signal from an array sensor is temporarily stored and applied along with the presen~
pixel value to the subtraction circuit which produces a difference between the present pixel value and the stored prior pixel value. The comparison circuit produces a defect signal if the difference exceeds a predetermined value. The subtraction and comparison circuits are operated in real time, with only timing delays for pipelining and synchronizakion purposes, at high frequencies to enable high resolution detection at high rates.
In a third aspect, the invention is summarized in a defect detection sy~tem having a plurality of defect detection circuits connected to a plurality of outputs, respectively, of a two-dimensional CCD array sensor operating in the TDI mode. A plurality of temporary memories, for example conventional first-in, first-out memories (FIFOs), are operated by the respective defect detecting circuits to store the defect pixel values detected in the video signal from the corresponding array output. An output circuit reads the temporary memories and transmits the defect pixel values.
Distributing the outputs of a TDI CCD array sensor to corresponding separate defect detecting circuits permits processing of pixels at lower frequencies while providing a greatly increased overall processing rate.
An object of the invention is to provide an economical and effective defect detection system employing an array sensor generating a video signal ' ':
.
2 ~ Pi! ~
wherein defects are detected at high resolutions and high rates.
Another object of the invention is to provide a hardware circuit which is simple and small so that it can be readily incorporated in a video camera ~or detecting object defects in a video signal of the object.
It is also an object of the invention to distribute the pixels from a TDI CCD array sensor in a camera to a plurality of defect detection circuits in the camera so that the video signals can be processed at relatively low frequencies.
one advantage of the invention is that the need ~or high frequency transmission of video signals from cameras to processing units is eliminated.
Another advantage of the invention is that the computer processing of quantities of video pixel information is greatly reduced.
A further advantage of the invention is that the detection of object defects at high resolutions can be made of objects, such as webs, moving at substantially higher rates than has been previously possible.
An additional feature of the invention is the provision of storing and transferring defect pixel coordinates along with defect pixel values.
A still additional feature of the invention is the provision of expanding initially stored coordinates by bits identifying the section of the object where the defect i5 found such as by the addition of bits used to multiplex channels in each camera and to multiplex cameras.
It is yet another feature of the invention that there is recognized that reliable defect detection and dynamic adjustment of acceptable range values can be performed by storage of a prior pixel value derived ~rom one or more prior pixels, subtraction of the stored prior pixel value from a present pixel value, comparison ' .
.
:
' ~ 5 ~ 3 of the dlfference with a predetermined value to determine a defect, and the update of the stored previous pixel value in the absence of the dstermination of a defect.
A further fea-ture of the invention is the use of half-full, *ull or other fullness measuring flags of temporary memories or FIFOs to indicate excessive defects and/or to indicate edges of webs under inspection.
A still further feature of the invention is the stopping of storing of defect pixels in a memory when the rate of accumulation of defect pixels exceeds the rate at which the defect pixels are read out.
Yet still a further feature of the invention is the provision for continuing the storage of end of line flags in a memory which has been shut down for excessive accumulation of unread stored defect pixels.
Other objects, advantages and features of the invention will be apparent from the following description of the preferred embodiment and the accompanying drawings.
Brief Description of Drawings Fig. 1 is a block diagram of a system for detecting defects in a moving web in accordance with the invention.
Fig. 2 is a block diagram of circuitry contained within a video camera of the system of Fig. 1.
Fig. 3 is diagrammatic view of a TDI CCD two-dimensional array sensor in the circuit of Fig. 2.
Fig. 4 is a functional block diagram of a defect detection circuit employed in the circuit of Fig. 2.
Fig. 5 is a detailed block diagram of electrical circuitry forming a defect detection circuit for one channel of the circuit of Fig. 2.
Figs. 6a and 6b, when joined with Fig. 6a on top and Fig. 6b on bottom, form a detailed electrical block diagram of circuitry incorporated in a field - 6 - 2~
programmable gate array unit (FPGA) in the circuit of Fig. 5.
Fig. 7 is a time graph of various signals in the circuitry of Figs. 5 and 6a.
Fig. 8 is a detailed block diagram of camera output circuitry in the circuit of Fig. 2.
Fig. 9 is a detailed block diagram of circuitry incorporated in a field programmable gate array unit tFPGA) in the circuit of Fig. 8.
Fig. 10 is a block diagram of circuitry included in a multiplexer and interface unit of Fig. 1.
Description of the Preferred Embodiment As shown in Fig. 1, one embodiment o~ a system and process for detectiny defects in an object, such as a moving web or plastic film 20, in accordance with the invention includes a conveyor 22 for directing the moving web through an inspection station having an illumination system 24 and a plurality of video cameras 30 mounted in a line perpendicular to the movement of the web. Lenses 32 of the cameras 30 are selected to view corresponding, slightly overlapping, sections along lines completely across the web. An encoder 44 driven by the conveyor 22 generates pulses at a frequency directly proportional to the speed of the conveyor 22.
The encoder pulses are appropriately shaped and amplified and applied over the line 45 to the cameras to synchronize the vertical scan with the web movement.
The cameras 30 themselves analyze the video signals generated from the viewed images to detect defects and produce data on each defect and its relative X-position or coordinate, that is, relative position from one side edge of the section viewed by each camera. Electrical cables 34 connect the cameras 30 to a multiplexer and interface unit 38 which combines the defect data for each transverse image line o~ the web 20 detected by all the cameras and then suitabIy passes the defect data over a cable 40 to an image processing unit 42 in a ~ : .
~ 7 ~ ~ 8 ~
central computer 46. Also the encoder pulses are applied by line ~5 to the multiplexer an~ interface unit 38.
The present system reduces the trans~er of video information from the video cameras to image processing units by a factor of looo or mo~e. ~emory requirements in the image processing units are similarly reduced to that necessary to store and process only the defect data. Requirements for image processing power, such as processing speed or multiple parallel processors, are likewise greatly reduced since the image data being processed in second stage processing is limited to defect data. This results in an economical web inspection system with greatly improved resolution o~
defects.
The conveyor 22, the illumination system 24, the en~oder 44, the computer 46, and the image processing unit 42 are conventional and can be similar to that described in the International Application No. ~0 2Q 91/14173 published on 19 September 1991. However, defect detection is performed by the cameras 30 and second stage processing 42 is simplified. The unit 42 contains a single processor (not shown) which receives thè defect data and suitably processes the data to produce an image of a defect on the monitor 50. The processed defect data is passed by the unit 42 to memory common with the central computer 46 which determines the type of defect in accordance with the algorithms disclosed in the International Application No. WO
91/14173. The image processing unit 42 also counts the image lines (one end of line flag is passed to the unit 42 for each image Iine of the web scanned by the cameras) to determine the Y-position, the position alung the length of the web, of each defect and passes this position information to the central computer 46.
The central computer 46 from the processed defect data, prints suitable defect information on printer 48, and can store the defect data in magnetic or other mass storage media (not ~hown). An input/output unit 52 in the central computer 46 operates address and data bus lines 54 connected to the unit 38 as well as to the illumination system 24 for providing control functions and for transmitting and receiving data.
Each of the cameras 30 includes, as illustrated in Fig. 2, a camera housing 60 in which are mounted a sensor printed circuit card or unit 62, four analog-to-digital conversion printed circuit cards or units 64, four defect detection printed circuit cards or units 66, a timing printed circuit card or unit 68, and a computer interface printed circuit card or unit 70. The sensor card 62, the analog-to-digital cards 64, and the timing card 68 are commercially available or contain conventional circuitry. The timing circuit 68, synchronized by the pulses on line 45 ~rom encoder 44, Fig. 1, operates and controls the sensor circuit 62, the analog-to-digital circuits 64 and the defect detection circuits 66.
The sensor card 62 includes a conventional TDI CCD
two-dimensional array sensor 80, Fig. 3, which has an array of sensing photosites arranged in columns 0 through 2047 and rows 0 through~96 upon which an image of a corresponding section of the moving web is 3Q projected by the camera lens. An electrical charge is generated and accumulated at each photosite proportional to the quantity of photons imPinging on each pho~osite.
Phase signals ~ 2, ~3 and ~4 generated in synchronism with encoder pulses on line 45, Fig. 2, simultaneously shift the rows of electrical charges in the direction o~
arrow 82 so that the charges move in correspondence to the movement of the image over the sensor array. In .
.
~3~
g this manner the charge for each image pixel is integrated or accumulated in the array as each image pixel moves in its corresponding column ~rom row 0 to row 96. When the integrated charges reach the row 96, the charges are transferred by transfer gates 84 to shift regi~ters 86. The shift registers 86 are operated in parallel by high frequency shift out signals to serially output charges between the shifts of rows of charges. The shift out signals serially trans~er the charges in the shift registers 86 to respective diffusion output regions 88 from which video analog voltage outputs ~0 are derived. Reset signals restore the diffusion regions between output pixel voltages.
Each of the eight outputs 90 corresponds to a respective section of each line of pixel voltages being outputted, such as pixels 255 to 0, 511 to 256, etc.
Thus the pixels of each line of an image portion viewed by a camera are distributed to the eight outputs 90.
The pixels of each section are outputted in reverse order so that the time displacement of each output pixel corresponds inversely to the X-position of the image pixel in the corresponding image section. Although the present sensor 80 outputs the pixels from each line image section in reverse order, the sensor 80 could be a sensor designed to output the pixels in forward order~
i.e. from 0 or 255, 256 to 511, etc.
Referxing back to Fig. 2, a pair of the outputs ~0 are applied to each analog to-digital circuit card 64 which contains two conventional analog-to-digital conversion circuits to generate respective output streams 92 of eight-bit parallel digital pixel signals which are applied to inputs on a respective defect detection circuit card 66. Clock or strobe signals on lines 94 hetween the circuit cards 64 and 66 are also supplied to the defect detection circuits.
Each of the defect detection circuit cards 66 contain two parallel circuits or channels for detecting ~ ' ~ . , .
2~830 ~4 defects in the respective digital pixel streams 92.
Each of these detection circuits include a data compression circuit (DCC) and a first-in first-out memory (FIF3). These circuits serve to eliminate non-relevant background pixels from the video digitalstreams except that one background pixel value is saved and transmitted along with each group of defect pixels, and one initial background pixel value is saved and transmitted along with an end of line flag (EOL) at the end of each scan line.
The function of the DCC and FIFO circuits is illustrated in Fig. 4 wherein digital pixel stream 92 is applied to one input of subtraction circuit 100 which has the digital output of a selector 102 applied to the second input of the subtraction circuit. One input of the selector is connected by two serially connected delay circuits 104 and 106 to the incoming digital pix~l stream 92 to normally apply a prior digital pixel value to the second input of the subtraction circuit. The difference output (Pj~Pjn) of the subtraction circuit 100, after passing through a delay 107, is compared with a predetermined upper difference limit value ~u in a comparator 108 and with a predetermined lower difference limit value ~~, in a comparator 110 wherein Pj is the present digital pixel value and Pin is a prior digital pixel value. The greater than output (~u > ~Pi~Pjn)) o~
the comparator 108 is connected by a delay circuit 112 to one input of a NAND gate 114 while the less than output (-~l, < (Pi~Pin)) of the comparator 110 is connected by a delay circuit 116 to the second input of the NAND
gate 114. In the absence of a defect, the greater than output of comparator 108 and the less than output o comparator 110 are both true (high) which generates a false (low) from gate 114 on DEFECT line 120. The DEFECT line 120 is connected to the control input oP
selector 102 to select the input from delay circuit 106 3~
to normally pass the digital pixel Pjn to the subtraction circuit in the absence of a defect.
The delays 106, 107, 112 and 116 are included to provide pipelining and maximum throughput of the circuit. The circuit would function properly without these delays 106, 107, 112 and 116 but would hava a lower maximum operating frequency.
The subtraction circuit loo, the comparators 108 and 110, and the NAND gate 114 provide a determination of whether the pixel ~alue is within a dynamic acceptable range (Pi.n ~ ~L) to (Pjn + ~u)- The acceptahle range is updated or dynamically changed by changing the value Pin to track normal gradual pixel changes.
A delay circuit 124 is connected between ~he output and the other input of the selector 102. When a pixel or pixels of a defect on digital input 92 are presented to subtraction circuit 100, the DEFECT line 120, after the appropriate delay, goes true (high) to operate selector 102 to supply the output of the delay circuit 124 to the second input of the subtraction circuit 100.
Thus~during a defect, the subtraction circuit subtracts a prior normal~digital pixel ~alue stored in the delay~
124 from the incoming pixel values. The selector 102 during a defect signal does not update the comparison or reference pixel value to a new Pin from delay 106 until the defect pixels are passed. It is noted that in the case of a defect, n in Pjn is increased dynamically by the number of clock pulses required to pass the defect and is returned to its initial value after the defect passes.
The clock or strobe input 94 operates the delay circuits 104, 106, 112, 116 and 124 to produce the proper synchronization of the circuit. Also the clock 94 steps a count-down counter 126 which is operated to indicate the relative X-position or coordinate o~ the pixel under test with respect to the beginning of the segment o~ line image being examined for de~ects; i~ the .
' 2 ~
sensor 80 is selected to output the pixels in ~orward order, the counter 1~6 is operated to count up to indicate the relative X-coordinate. When the DEFECT
line 120 goes true the input control of FIFO 130 is operated to store the defect pixel value and its X
coordinate from counter 126 in the FIFO. The stored defect pixel data is subsequently read out of the FIFOs by the computer interface circuit (CIC) 70, Fig. 2l and transferred to the corresponding secondary processing unit for furthex analysis.
The circuitry forming the DCC and ~IF0 for one channel on one of the cards 66 is shown in Fig. 5 and includes a field programmable gate array chip (FPGA) 136, two nine-bit conventional FIFO chips 130a and 130b, and two four-bit conventional counter chips 126a and 126b. Conveniently, each of the cards 66 are manufactured with identical circuitry and have jumpers (not shown~ for enabling channel selection in a conventional manner during camera assembly.
Figs. 6a and 6b show a hardware design of suitable DCC circuitry for being incorporated in an field programmable gate array (FPGA) such that available from Altera Corporation, San Jose, California, USA. Because of design limitations andÇor optimizing design decisions, the subtraction circuit 100 of Fig. 4 is incorporated in a pair of cascaded four-bit ALU circuits lOOa and lOOb (still labeled as subtract circuits) in Fig. 6a, and the selector circuit 102 of Fig. 4 is incorporated is a pair of four-bit multiplexer circuits 102a and 102b (still labeled as selectors) formed as Boolean units. The delays 104, 106, 107 and 124 of Fig.
of pixel values in the video signal, for example, by a ratio of 1:1000 or less of defect pixels to total pixels.
In a second aspect, the invention is summarized in a defect detection system having a subtraction circuit along with a comparison circuit. A prior pixel value derived from one or more pixels prior to a present pixel value in a video signal from an array sensor is temporarily stored and applied along with the presen~
pixel value to the subtraction circuit which produces a difference between the present pixel value and the stored prior pixel value. The comparison circuit produces a defect signal if the difference exceeds a predetermined value. The subtraction and comparison circuits are operated in real time, with only timing delays for pipelining and synchronizakion purposes, at high frequencies to enable high resolution detection at high rates.
In a third aspect, the invention is summarized in a defect detection sy~tem having a plurality of defect detection circuits connected to a plurality of outputs, respectively, of a two-dimensional CCD array sensor operating in the TDI mode. A plurality of temporary memories, for example conventional first-in, first-out memories (FIFOs), are operated by the respective defect detecting circuits to store the defect pixel values detected in the video signal from the corresponding array output. An output circuit reads the temporary memories and transmits the defect pixel values.
Distributing the outputs of a TDI CCD array sensor to corresponding separate defect detecting circuits permits processing of pixels at lower frequencies while providing a greatly increased overall processing rate.
An object of the invention is to provide an economical and effective defect detection system employing an array sensor generating a video signal ' ':
.
2 ~ Pi! ~
wherein defects are detected at high resolutions and high rates.
Another object of the invention is to provide a hardware circuit which is simple and small so that it can be readily incorporated in a video camera ~or detecting object defects in a video signal of the object.
It is also an object of the invention to distribute the pixels from a TDI CCD array sensor in a camera to a plurality of defect detection circuits in the camera so that the video signals can be processed at relatively low frequencies.
one advantage of the invention is that the need ~or high frequency transmission of video signals from cameras to processing units is eliminated.
Another advantage of the invention is that the computer processing of quantities of video pixel information is greatly reduced.
A further advantage of the invention is that the detection of object defects at high resolutions can be made of objects, such as webs, moving at substantially higher rates than has been previously possible.
An additional feature of the invention is the provision of storing and transferring defect pixel coordinates along with defect pixel values.
A still additional feature of the invention is the provision of expanding initially stored coordinates by bits identifying the section of the object where the defect i5 found such as by the addition of bits used to multiplex channels in each camera and to multiplex cameras.
It is yet another feature of the invention that there is recognized that reliable defect detection and dynamic adjustment of acceptable range values can be performed by storage of a prior pixel value derived ~rom one or more prior pixels, subtraction of the stored prior pixel value from a present pixel value, comparison ' .
.
:
' ~ 5 ~ 3 of the dlfference with a predetermined value to determine a defect, and the update of the stored previous pixel value in the absence of the dstermination of a defect.
A further fea-ture of the invention is the use of half-full, *ull or other fullness measuring flags of temporary memories or FIFOs to indicate excessive defects and/or to indicate edges of webs under inspection.
A still further feature of the invention is the stopping of storing of defect pixels in a memory when the rate of accumulation of defect pixels exceeds the rate at which the defect pixels are read out.
Yet still a further feature of the invention is the provision for continuing the storage of end of line flags in a memory which has been shut down for excessive accumulation of unread stored defect pixels.
Other objects, advantages and features of the invention will be apparent from the following description of the preferred embodiment and the accompanying drawings.
Brief Description of Drawings Fig. 1 is a block diagram of a system for detecting defects in a moving web in accordance with the invention.
Fig. 2 is a block diagram of circuitry contained within a video camera of the system of Fig. 1.
Fig. 3 is diagrammatic view of a TDI CCD two-dimensional array sensor in the circuit of Fig. 2.
Fig. 4 is a functional block diagram of a defect detection circuit employed in the circuit of Fig. 2.
Fig. 5 is a detailed block diagram of electrical circuitry forming a defect detection circuit for one channel of the circuit of Fig. 2.
Figs. 6a and 6b, when joined with Fig. 6a on top and Fig. 6b on bottom, form a detailed electrical block diagram of circuitry incorporated in a field - 6 - 2~
programmable gate array unit (FPGA) in the circuit of Fig. 5.
Fig. 7 is a time graph of various signals in the circuitry of Figs. 5 and 6a.
Fig. 8 is a detailed block diagram of camera output circuitry in the circuit of Fig. 2.
Fig. 9 is a detailed block diagram of circuitry incorporated in a field programmable gate array unit tFPGA) in the circuit of Fig. 8.
Fig. 10 is a block diagram of circuitry included in a multiplexer and interface unit of Fig. 1.
Description of the Preferred Embodiment As shown in Fig. 1, one embodiment o~ a system and process for detectiny defects in an object, such as a moving web or plastic film 20, in accordance with the invention includes a conveyor 22 for directing the moving web through an inspection station having an illumination system 24 and a plurality of video cameras 30 mounted in a line perpendicular to the movement of the web. Lenses 32 of the cameras 30 are selected to view corresponding, slightly overlapping, sections along lines completely across the web. An encoder 44 driven by the conveyor 22 generates pulses at a frequency directly proportional to the speed of the conveyor 22.
The encoder pulses are appropriately shaped and amplified and applied over the line 45 to the cameras to synchronize the vertical scan with the web movement.
The cameras 30 themselves analyze the video signals generated from the viewed images to detect defects and produce data on each defect and its relative X-position or coordinate, that is, relative position from one side edge of the section viewed by each camera. Electrical cables 34 connect the cameras 30 to a multiplexer and interface unit 38 which combines the defect data for each transverse image line o~ the web 20 detected by all the cameras and then suitabIy passes the defect data over a cable 40 to an image processing unit 42 in a ~ : .
~ 7 ~ ~ 8 ~
central computer 46. Also the encoder pulses are applied by line ~5 to the multiplexer an~ interface unit 38.
The present system reduces the trans~er of video information from the video cameras to image processing units by a factor of looo or mo~e. ~emory requirements in the image processing units are similarly reduced to that necessary to store and process only the defect data. Requirements for image processing power, such as processing speed or multiple parallel processors, are likewise greatly reduced since the image data being processed in second stage processing is limited to defect data. This results in an economical web inspection system with greatly improved resolution o~
defects.
The conveyor 22, the illumination system 24, the en~oder 44, the computer 46, and the image processing unit 42 are conventional and can be similar to that described in the International Application No. ~0 2Q 91/14173 published on 19 September 1991. However, defect detection is performed by the cameras 30 and second stage processing 42 is simplified. The unit 42 contains a single processor (not shown) which receives thè defect data and suitably processes the data to produce an image of a defect on the monitor 50. The processed defect data is passed by the unit 42 to memory common with the central computer 46 which determines the type of defect in accordance with the algorithms disclosed in the International Application No. WO
91/14173. The image processing unit 42 also counts the image lines (one end of line flag is passed to the unit 42 for each image Iine of the web scanned by the cameras) to determine the Y-position, the position alung the length of the web, of each defect and passes this position information to the central computer 46.
The central computer 46 from the processed defect data, prints suitable defect information on printer 48, and can store the defect data in magnetic or other mass storage media (not ~hown). An input/output unit 52 in the central computer 46 operates address and data bus lines 54 connected to the unit 38 as well as to the illumination system 24 for providing control functions and for transmitting and receiving data.
Each of the cameras 30 includes, as illustrated in Fig. 2, a camera housing 60 in which are mounted a sensor printed circuit card or unit 62, four analog-to-digital conversion printed circuit cards or units 64, four defect detection printed circuit cards or units 66, a timing printed circuit card or unit 68, and a computer interface printed circuit card or unit 70. The sensor card 62, the analog-to-digital cards 64, and the timing card 68 are commercially available or contain conventional circuitry. The timing circuit 68, synchronized by the pulses on line 45 ~rom encoder 44, Fig. 1, operates and controls the sensor circuit 62, the analog-to-digital circuits 64 and the defect detection circuits 66.
The sensor card 62 includes a conventional TDI CCD
two-dimensional array sensor 80, Fig. 3, which has an array of sensing photosites arranged in columns 0 through 2047 and rows 0 through~96 upon which an image of a corresponding section of the moving web is 3Q projected by the camera lens. An electrical charge is generated and accumulated at each photosite proportional to the quantity of photons imPinging on each pho~osite.
Phase signals ~ 2, ~3 and ~4 generated in synchronism with encoder pulses on line 45, Fig. 2, simultaneously shift the rows of electrical charges in the direction o~
arrow 82 so that the charges move in correspondence to the movement of the image over the sensor array. In .
.
~3~
g this manner the charge for each image pixel is integrated or accumulated in the array as each image pixel moves in its corresponding column ~rom row 0 to row 96. When the integrated charges reach the row 96, the charges are transferred by transfer gates 84 to shift regi~ters 86. The shift registers 86 are operated in parallel by high frequency shift out signals to serially output charges between the shifts of rows of charges. The shift out signals serially trans~er the charges in the shift registers 86 to respective diffusion output regions 88 from which video analog voltage outputs ~0 are derived. Reset signals restore the diffusion regions between output pixel voltages.
Each of the eight outputs 90 corresponds to a respective section of each line of pixel voltages being outputted, such as pixels 255 to 0, 511 to 256, etc.
Thus the pixels of each line of an image portion viewed by a camera are distributed to the eight outputs 90.
The pixels of each section are outputted in reverse order so that the time displacement of each output pixel corresponds inversely to the X-position of the image pixel in the corresponding image section. Although the present sensor 80 outputs the pixels from each line image section in reverse order, the sensor 80 could be a sensor designed to output the pixels in forward order~
i.e. from 0 or 255, 256 to 511, etc.
Referxing back to Fig. 2, a pair of the outputs ~0 are applied to each analog to-digital circuit card 64 which contains two conventional analog-to-digital conversion circuits to generate respective output streams 92 of eight-bit parallel digital pixel signals which are applied to inputs on a respective defect detection circuit card 66. Clock or strobe signals on lines 94 hetween the circuit cards 64 and 66 are also supplied to the defect detection circuits.
Each of the defect detection circuit cards 66 contain two parallel circuits or channels for detecting ~ ' ~ . , .
2~830 ~4 defects in the respective digital pixel streams 92.
Each of these detection circuits include a data compression circuit (DCC) and a first-in first-out memory (FIF3). These circuits serve to eliminate non-relevant background pixels from the video digitalstreams except that one background pixel value is saved and transmitted along with each group of defect pixels, and one initial background pixel value is saved and transmitted along with an end of line flag (EOL) at the end of each scan line.
The function of the DCC and FIFO circuits is illustrated in Fig. 4 wherein digital pixel stream 92 is applied to one input of subtraction circuit 100 which has the digital output of a selector 102 applied to the second input of the subtraction circuit. One input of the selector is connected by two serially connected delay circuits 104 and 106 to the incoming digital pix~l stream 92 to normally apply a prior digital pixel value to the second input of the subtraction circuit. The difference output (Pj~Pjn) of the subtraction circuit 100, after passing through a delay 107, is compared with a predetermined upper difference limit value ~u in a comparator 108 and with a predetermined lower difference limit value ~~, in a comparator 110 wherein Pj is the present digital pixel value and Pin is a prior digital pixel value. The greater than output (~u > ~Pi~Pjn)) o~
the comparator 108 is connected by a delay circuit 112 to one input of a NAND gate 114 while the less than output (-~l, < (Pi~Pin)) of the comparator 110 is connected by a delay circuit 116 to the second input of the NAND
gate 114. In the absence of a defect, the greater than output of comparator 108 and the less than output o comparator 110 are both true (high) which generates a false (low) from gate 114 on DEFECT line 120. The DEFECT line 120 is connected to the control input oP
selector 102 to select the input from delay circuit 106 3~
to normally pass the digital pixel Pjn to the subtraction circuit in the absence of a defect.
The delays 106, 107, 112 and 116 are included to provide pipelining and maximum throughput of the circuit. The circuit would function properly without these delays 106, 107, 112 and 116 but would hava a lower maximum operating frequency.
The subtraction circuit loo, the comparators 108 and 110, and the NAND gate 114 provide a determination of whether the pixel ~alue is within a dynamic acceptable range (Pi.n ~ ~L) to (Pjn + ~u)- The acceptahle range is updated or dynamically changed by changing the value Pin to track normal gradual pixel changes.
A delay circuit 124 is connected between ~he output and the other input of the selector 102. When a pixel or pixels of a defect on digital input 92 are presented to subtraction circuit 100, the DEFECT line 120, after the appropriate delay, goes true (high) to operate selector 102 to supply the output of the delay circuit 124 to the second input of the subtraction circuit 100.
Thus~during a defect, the subtraction circuit subtracts a prior normal~digital pixel ~alue stored in the delay~
124 from the incoming pixel values. The selector 102 during a defect signal does not update the comparison or reference pixel value to a new Pin from delay 106 until the defect pixels are passed. It is noted that in the case of a defect, n in Pjn is increased dynamically by the number of clock pulses required to pass the defect and is returned to its initial value after the defect passes.
The clock or strobe input 94 operates the delay circuits 104, 106, 112, 116 and 124 to produce the proper synchronization of the circuit. Also the clock 94 steps a count-down counter 126 which is operated to indicate the relative X-position or coordinate o~ the pixel under test with respect to the beginning of the segment o~ line image being examined for de~ects; i~ the .
' 2 ~
sensor 80 is selected to output the pixels in ~orward order, the counter 1~6 is operated to count up to indicate the relative X-coordinate. When the DEFECT
line 120 goes true the input control of FIFO 130 is operated to store the defect pixel value and its X
coordinate from counter 126 in the FIFO. The stored defect pixel data is subsequently read out of the FIFOs by the computer interface circuit (CIC) 70, Fig. 2l and transferred to the corresponding secondary processing unit for furthex analysis.
The circuitry forming the DCC and ~IF0 for one channel on one of the cards 66 is shown in Fig. 5 and includes a field programmable gate array chip (FPGA) 136, two nine-bit conventional FIFO chips 130a and 130b, and two four-bit conventional counter chips 126a and 126b. Conveniently, each of the cards 66 are manufactured with identical circuitry and have jumpers (not shown~ for enabling channel selection in a conventional manner during camera assembly.
Figs. 6a and 6b show a hardware design of suitable DCC circuitry for being incorporated in an field programmable gate array (FPGA) such that available from Altera Corporation, San Jose, California, USA. Because of design limitations andÇor optimizing design decisions, the subtraction circuit 100 of Fig. 4 is incorporated in a pair of cascaded four-bit ALU circuits lOOa and lOOb (still labeled as subtract circuits) in Fig. 6a, and the selector circuit 102 of Fig. 4 is incorporated is a pair of four-bit multiplexer circuits 102a and 102b (still labeled as selectors) formed as Boolean units. The delays 104, 106, 107 and 124 of Fig.
4 are formed by eight-bit latch circuits 104, 106, 107 and 124 of Fig. 6a. The delays 112 and 116 of Fig. 4 are incorporated in respective stages of a nine bit latch circuit 136 in Fig. 6a. An additional delay is interposed between the output of the NAND gate 114 and the DEFECT signal line 120 by a stage of the latch 136 while an additional eight-bit delay latch 13~ is interposed between the output of latch 106 and the inputs of selectors 102a and 102b, Eight bit latches 140, 142 and 144 have data inputs connected to a system bus 146 which shares lines RB/SB
with the eight-bit gray value portion o~ output data from the FIF0 130a to the computer interface circuit 70 of Fig. 2. Clock inputs of latches 140, 142 and 144 are connected to respective control lines 148 (PHIL), 150 (MWL) and 152 (TINITL) from the computer interface circuit for storing operating values transferred over the system bus SB from the central computer 46, Fig~ 1, through the interface unit 38. The value ~u is receivPd and applied by latch 140 to the compare circuit 108.
The value ~L iS received and applied by latch circuit 142 to the compare circuit 110. An initial pixel value is received in the latch 144 for flushing prior pixel values in the pipeline and for presenting the initial prior pixel value upon start of the detection of line pixel values. The strobe input 94 is inverted by an inverter 168 to form a clock signal CLK3 which operates various latches in Fig. 6a and operates the counter 126a in Fig. 5.
An encoder pulse input 45 (LV~L) from the~timing circuit 68 of Fig. 2 is high (see also Fig. 7j when data on pixel input 92 is the actual digital value of a pixel in a video line being outputted from the analog-to-digital converter units 64. The input LVAL is delayed by two stages of a latch circuit 182 operated by the strobe 94 to operate four-bit selectors 184a and 184b which have first inputs connected to the output of the latch 144 and second inputs connected to an output of an eight-bit latch 185 receiving the input pixels 92. The outputs of the selectors 184a and 184b are connected to the minuend inputs of the subtracting circuits lOOa and lOOb as well AS to the input of latch 104. The subtrahend inputs of the subtracting circuits lOOa and .
~, . . . . .
!
lOOb are connected to the outputs of the selectors 102a and 102b.
The carry or borrow output 186 of the unit lOOa is a ninth bit delayed by the latch 107 and a stage of latch 136 to enable a NAND gate 188 via inverter 190 when the carry signal 186 is low indicating that the minuend P; is equal to or greater than the subtrahend P~n and to enable gate 192 when the carry signal is high indicating a negative difference value. This gating is necessary since negative differences outputted from subtract circuits lOOa and lOOb are twos complements.
When the gate 188 is enabled, the greater than output of the upper limit comparator 108 is passed through a stage of the latch 136 and the gate 18~ so that the gate 114 passes the result from the upper limit comparator as the defect signal 120. When the gate 192 is enabled, tha less than output of the lower limit comparator llO is passed through the latch 136 and the gate 192 so that the gate 114 passes the result from the lower limit comparator as the defect signal 120.
The use of a prior pixel value as a reference value for a dynamic range of acceptable values in the circuit of Figs. 5 and 6a is based upon a finding that normal pixel values vary very slowly in plastic films being analyzed. However, different testing conditions, such as larger induced noise levels, could cause excessive errors in defect detection. Under such different conditions, errors in defect detection can be reduced by the employment of a pixel averaging circuit (not shown) or a filtering circuit (not shown) such as a finite impulse response (FIR) filter (not shown) or an infinite impulse response (IIR) filter (not shown) in series with latches 104 and 106 so that the prior pixel value presented to the selector 102 is an average or filt~red value derived from one or more prior pixels.
The delayed LVAL signal from latch 182 is further delayed by three staqes of latch 136 and applied by - 15 ~ r~
inverter 198 as the signal STENDEL/ to the CLR input of the counters 125a and 126b. In the absence of incoming pixels, i.e. LVAL being low, the counters 126a and 126b are reset to zero. When the pixels are outputted from the sensor in reverse order as described herein, the outputs of the counters 126a and 126b are to be inverted to correspond to the X-position of the incoming pixels;
due to hardware limitations, this inversion is performed by software in the image processing unit 42 rather than by count-down counters or inverters in the camera. If the incoming pixels were not reversed, such inversion or count-down counter is unnecessary.
The defect signal 120 is applied by an inverter 202 to one input of a NAND gate 204 which enables a NAND
gate 206 to pass the strobe signal 94 to output FCLK.
The output FCLK is applied to one input of a NAND gate 208, Fig.6b, which, when enabled, generates a write signal W/ to operate the FIFOs 130a and 130b, Fig. 5, to store the value of the defect pixel on bus 132 from latch 138 in FPGA 136 and to store the position or X-coordinate on bus 128 from the counters 126a and 126b.
Additionally the defect signal is further delayed by another stage of the latch 136 and applied by an inverter 210 to a second input of NAND gate 204 so that the position and value of the first normal pixel following a series of one or more defect pixels is stored in the FIFOs 130a and 130b. The value and position of this normal pixel are transmitted along with the dPfect pixel data from the cameras to the second stage processing apparatus to enable the further processing of this data.
The thrice delayed signal LVAL from latch 136 is applied by inverter 212 to an output EOLBIT which is applied as one bit input to the FIFO 130b. The twice delayed LVAL signal from latch 136 is inverted by an inverter 214 and applied to one input of a NAMD gate 216 which receives the thrice delayed LVAL signal on its ., ~ .
- 16 - ~ ~83~
other input. The output of the NAND yate 216 is inverted by an inverter 220, delayed by another st~ge in latch 136, inverted by an inverter 222, and applied to a third input of NAND gate 204 to store a high EOLBIT
signal in the FIFO 13Ob along with the position zero and the initial value from latch 144 after the last incoming pixel value of a line section has been tested for a defect value. During the evaluation of pixels in a line section, the EOLBIT signal is low so that the corresponding FIFO stored bit in each defect pixel record is low.
The computer interface has a defect override line 224 (DEFOVRD/ ) connected to a fourth input of the NAND
gate 204 to output all the pixels passing to output 132.
The line 224 is normally high and is rendered low when it is desired to look at pixels or values other than the defect pixels, background pixels, and end of line bits normaIly stored in the FIFOs.
An example of operation of the circuit of Fig. 6a is illustrated in Fig. 7. The upper waveform (CLOCK) represents the clock signal CLK3 distributed to various portions of the circuit. It is noted that Fig. 7 illustrates the relative timing of various circuit portions, but Fig. 7 is not in sufficient detail to accurately refle~t the relative phase shift inherent in the various circuit components of Fig.6a. The values MINUEND, SUBTRAHEND, DIFFERENCE and COUNTER ars set forth in Fig. 7 as decimal numbers, but in the circuit of Fig. 6a these values are presented as parallel eight-bit binary signals. The row DIFFERENCE containsnegative decimal numbers, but in the circuitry of Fig.
6a, negative numbers are the twos compl0ment. The row PIXEL NUMBER in Fig. 7 is inserted for reference purposes to identify the MINUEND image pixels and does not represent any values or signals in Fig. 6a. The COUNTER values represent inverted outputs o~ counters 126a and 126b of Fig. 5.
While LVAL is low, the initial value, one hundred twenty, stored in latch 144 is applied to the minuend inputs of the subtracting circuits lOOa and lOOb. Since this value will have been flushed through the latches 104, 106 and 138, the subtrahend is also equal to one hundred twenty and the difference is zero. When LVAL
goes high indicating that line pixel values are now being applied to latch 185, the selectors 184a and 184b, after a delay, are operated to apply the incoming pixels to the minuend input of the subtracting circuits lOOa and lOOb. The first line image pixel, number 255, has a value of one hundred twenty-one, while pixels, numbers 254-252, have values of one hundred twenty-two. These values do not appear in the subtrahend until three clock periods later so difference values of plus one and plus two are generated. The value ~u is set at plus three and the value -~L iS set at minus three, i.e. llllllO1.
Since plus two is not greater than plus three, the compare high output of the upper limit comparator 108 remains low. The compare low output of lower limit comparator 110 does go high, but the carry signal on line 186 remains low so the gate 192 is disabled to block passage of the compare~low signal to the defect gate 114.
Pixels, numbers 251-248, are defect pixels with values of one hundred sixty-four. Since the difference of plus forty-two is greater than plus three, the compare high output goes high to produce a high defect signal DEFECT. The high defect signal enables the FCLK
signal so that the values of pixels 251-248 along with the pixel X-positions from counters 126a and 126b are stored in the FIFOs 130a and 130b. The high defect signal DEFECT on line 120 also operates the selectors 102a and 102b to prevent the defect values, one hundred sixty-four, being applied to the subtrahend. Rather the subtrahend is held at the last subtrahend value stored in latch 124 before the defect signal, i.e. one hundred , ' -- 18 - ~ 8~ ~L~
twenty-two. In the clock period following the end o~
the defect signal DEFECT, the next normal pixel value and its coordinate are stored in the FIFOs due to the delayed defect signal through inverter 210 to operate gate 204.
Pixels, numbers 9 and 8, have values, one hundred twenty-two, applied to the minuend input which are less than the value, one hundred twenty-three, applied from a prior normal pixel to the subtrahend input. This results in a difference of negative one (1111111~ which is not less than the value -~L (11111101) SO th2 compare low output of the lower limit comparator 110 is low.
The carry signal is high to disable the gate 188 and to prevent the compare high signal from operating the defect gate 114.
Pixels, numbers 7-5, have values, thirty-one, which produce a difference of negative ninety-two (10100100) which is less than negative three (11111101) to produce a high defect signal DEFECT on line 120. This results in the storing of the values of pixels, numbers 7-4, along with their coordinates in the FIFOs.
When LVAL goes low marking the end of line pixel input, the coincidence of inverted twice delayed LVAL
and thrice delayed LVAL operates gate 216 which afker a further delay operates gate 204 to produce a pulse in FCLK to store a further record in the FIFOs 130a and 130b. This record contains the high EOLBIT signal to indicate the end of a line. Also this end of line record contains the initial value from latch 144 which will have been flushed through the pipeline.
The FIFOs 13Oa and 13Ob of Fig. 5 have outputs HF/
and EF/ which, when low, indicate over half full and empty conditions, respectively, of the FIFOs. The half full signal HF/ is applied by inverker 230 in Fig. 6b to the clock input of a flip-flop 232. The inverted output of the flip-flop 232 generates an output ALARM/ which is high in the absence of the half full signal and is ., .
,:
.
applied by OR gate 234 to an inpuk o~ the NAND gate 208.
When the half full signal HF/ goes low, the flip-flop 232 is operated to render the output ALARM/ low and disable the gate ~08. This prevents data of any further defect signals from being written to the FIFOs o~ the channel producing the alarm signal. However, the end o~
line signal EoLsIT is applied to the second input of OR
gate 234 so that end of line records continue to be written in the FIFOs. The size of the FIFOs 130a and 130b, for example 2048 x 9 bit FIFOs, is selected to allow storage of a normal maximum number of defects.
When defect pixels are detected at a rate greater than the rate at which the pixels are read out, the FIFOs over a period of time become half full to shut down storage of further defect pixels.
The alarm signals indicating shutdown of a channel have an additional function, namely that of indicating the edge of the web being examined for defects. The channels viewing image sections outside or straddling the edges of the web detect large numbers of de~ect pixels which quickly produce an alarm signal and shut down the defect detection of those channels~ The alarm signal can be turned off by either a RESTART/ s.ignal or a RST/ signal applied to a respective input of an AND
gate 236 which clears the flip-flop 232. The RST/
signal also resets the latches 182 and 183 in Fig. ~a, operates a flip~flop 238 in Fig. 6b to disable the write signal W/ for the remaining portion of a strobe pulse cycle after termination of the RST/ signal, and is passed of line RS/ to reset the FIFOs 130a and 130b of Fig. 5.
As illustrated in Fig. 2, the computer interface circuit 70, connec~s the defect detection circuits 66 to the cable 34 to the multiplexer and interface unit 38, Fig. 1. The circuit 70, as shown in Fig. 8, includes RS-422 type receivers 240 and 242 and RS-422 type transmitters 244, 246, 248 and 250 connected to lines in .
.
.. ~ .
.
2 ~ 7 ~
the cable 34. The receivers 240 receive control signals RST/, PHIL, TINITL, DEFOVRD/, RESTART/, and MWL
which are passed to the defect detection circuits 66, and receive a control signal SSPUW/ which, along with the signal RST/, is applied to a FPGA 252 controlling the multiplexed reading of the eight defect detectisn channels in circuits 66. The receivers 242 receive eight-bit data signals which are passed to the system bus (RB/SB) 146 when SSPUW/ is low and CAMERAW/ is high, and the transmitters 246 transmit signals on bus RB/SB
146 from FIFO 130a when SSPUW/ is high and CAMERAW/ is low. Transmitters 244 transmit control signals CAMERAW/
and DVALID from the FPGA 252 along with an alarm data signal ALARM Erom NAND gate 254 which has inputs connected to all eight of the alarm outputs of the defect detection circuits. Additionally the transmitters 244 transmit the end of line bit EOL and three bits X1, X2 and X3 which identify which one of the eight data detection channels is being read. The X-coordinate data in the defect records being read are transmitted from bus CB connected to the data outputs of the FIFOs 130b in the eight defect detection circuits when the signal CAMERAW/ is low. When CAMERAW~ is high, the transmitters 248 pass the alarm signals ALARMlj -ALARM8/ from the eight defect detection channels tv the RS-422 lines in cable 34 shared with the outputs of transmitters 250.
In the FPGA 252 as shown in Fig. 9, the outputs Xl, X2 and X3 of a counter are dacoded by conventional decoding circuitry 262 which through NAN~ gates 264 enabled when the signal SSPUW/ is high generates a corresponding one of the signals REN1/ - REN8/.
Referring back to Fig. 6b, the RENX input in each de~ect detection circuit is connected by a jumper (not shown) to a corresponding one of the eight lines RENl/ - REN8/, Fig. 9, to provide for multiplexing the reading of the defect data in the FIFOs 130a and 130b of the eight - 21 - ~ 7~
defect detection circuits. The input RENX/ is applied by inverter 272 to one input of an AND gate 274 which has a second input receiving the master clock signal CLK
from the timing circuit 68, Fig. 2. Third and fourth inputs of the AND yate 274 are connected to the line RST/ and the output of the flip-flop 238 to disable reading during resetting of the circuit. The output of the AND gate 274 is passed throuyh a NAND gate 276 to the line R/ to read records from the FIFOs 130a and 130b onto the buses CB and RB/SB until an EOL bit is produced. The second input of the NAND gate 276 is connected to an output of a flip-flop 278 which is normally high due to the strobe signal applied through inverter 280 to the clock input of the flip-flop 278. A
FIFO empty signal line EF/ from FIFO 130b is connected by inverter 282 to the clear input of a flip-flop 284 which, in the absence of the empty signal, is held clear to prevent its inverted output from operating the clear input of flip-flop 278. However when EF/ goes low, the flip-flop 28~ goes high on the next pulse from NAND gate 276 to clear flip-flop 278 and to disable gate 276 and prevent the reading of empty FIFOs. When EF/ goes high indicating that the FIFOs are not empty, the flip-flop 284 is cleared to remove the clear signal from flip-flop 278 and permit flip-flop 278 to be set high by the next strobe pulse.
The RENX/ signal is also applied by an inverter 286 to an enable input of a tristate buffer 288 which then passes the Rj signal from an inverter 290 to the signal line DVAL which indicates the presence of valid defect data~on the CB and RB/SB buses.
The EOL bit from each FIFO 13Ob is applied to the clock enable input of the counter 260 which is then stepped by the DVAL pulse on its clock input to advance the count in the counter 260. This advances control tv the next REN2/ - REN~/ control line until the last channel has been read. An AND qate 294 has inputs from ~ ~ .
.
.
- 22 ~
all the signals REN1/ - REN8/ to produce the CAMERAW/
signal during readout of the defect data uncoversd in an image line. An inverter 296 applies the output of the AND gate 294 to one input of an A~D gate 298 to enable the gate 298 to pass the signal DVAL through inverters 302 and 304 as the signal DVALID. A third input to the AND gate 298 is received from a NAND gate 306 which has inputs connected to the line REN8/ and to the line ~OL
to disable the gate 298 during each EOL signal except when the circuit 252 i5 reading the last deect detection channel in the camera. Thus the records with high EOL bits from the FIFOs are not recognized by the multiplexer and interface unit 38 as being valid data except for the records with EOL signals from the last camera channel and only one record with an EOL bit high is received as valid data by the multiplexer and interface unit 38 for each image line read out of a camera. If an image line analyzed by the camera contains no pixels of a defect, only a singIe record with an end of line bit high is transmitted from the camera to the multiplexer and interface unit 38.
Each pixel record in the FIFOs 130a and 130b includes on end of line bit (EOL), eight bits of pixel coordinate, and eight bits of pixel intensity. Three additional coordinate bits, identifying the channel in the camera, are added to each data record in transmission to the multiplexer and interface unit 38 to bring the total number of bits to twenty during this transmission. The multiplexer and interface unit 38 adds ~our more coordinate bits identifying the camera from which the data record is read to increase the record bits to twenty-four total bits transmitted from the unit 38 to the image processing unit 42.
As illustrated in Fig. l0, the multiplexer and interface unit 38 contains a central control 320 connected by individual control and data lines to a plurality of substantially similar printed circuit units 322 corresponding to the plurality of ~ameras 30, Fig.
1. Each of the units 322 include RS-422 type transmitters 322 and receivers 3~4 connected to appropriate lines in the corresponding cable 34. The system is under the general control of the computer 46, Fig. 1, through the bus 54 connected to input/output circuit 328. Bus 330 connected to the circuit 328 includes RST/ and RESTART/ lines (not ~hown~ connected to control 320 as well as to each of the units 322; a RST/ signal resets the entire system while a RESTART/
signal clears the alarm flags caused by the HF/ signals from the FIFOs. The bus 330 also includes PHIL, MWL, TINITL and DEFOVRD lines (not shown) along with camera address lines Po-P3 (not shown) connected to the control circuit 320. Buses 332 each including PHIL, MWL, TINITL
and DEFOVRD lines (not shown) are connected from the control circuit 320 to the transmitters in the respective units 322. The bus 330 further contains eight SB data lines (not shown) which are connected to transmitters in block 324 controlled by a CAMERAWX/
signal on line 334 from the control circuit 320. In response to a PHIL, MWL, TINITL or DEFOVRD signal over bus~330 from the computer, the control unit 320 sends the corresponding PHIL, MWL, TINITL or DEFOVRD over the bus 332 to the unit 322 addressed by the signals P0-P3 so as to enable the computer to set the values in latches 140, 142 and 144, Fig. 6a, in the selected cameras to the values SB on the system bus 146.
Each of the units 322 includes three FIFOs 336 to which the twenty incoming data lines in bus 338 from receivers 326 are connected for receiviny and temporarily storing the defect data records from the corresponding camera including an EOL flag bit, eight pixel intensity value bits RB and eleven X-coordinate bits CB. A bus 340 including DVALID, CAMERAWX/ and ALARM lines (not shown) is connected from the receivers 326 to the control circuit 320 which in response to the , . . .
~3~
DVALID and CAMEP~WX/ signals applies write signals to a line W/ to stored each incoming record in the FIFOs 336.
The control circuit also contains circuitry, similar to that of Fig. 6b, connected to a line HF/ from the FIFOs 336 for blocking the write signals, except when a signal is on the line EOL, in response to the FIFOs becoming half full to indicate excessive defects being detected by the corresponding camera. Production of write signals can be restarted by a RESTART/ or a RST/ signal from the computer.
Data outputs of the FIFOs 336 in all the units 322 are connected to a common twenty line bus 344 to inputs of FIFOs 346 which have four additional inputs connected to a bus 348 from the control unit 320 carrying camera address signals. In response to the absence of an EF/
or empty signal from the FIFOs 336, the control unit 320 generates read signals R/ applied to the FIFOs 336 in one selected unit 322 to output defect data records from that selected unit 322 until a record contains an EOL
flag. Simultaneously, the control unit 320 generates write signals on a line W/ to the FIFOs 346~to store the records being read from the selected unit 322 into the FIFOs 346. The control unit 320 contains circuitry similar to Fig. 9 responsive to the EOL signal on bus 344 for suppressing the write signal W/ to FIFOs 346, except when the twelfth unit 322 is being read. Thus records of defects in each image line extending across the view of all twelve cameras are assembled in the FIFOs 246 with an end record having the EOL bit flag from the tweIfth camera. If an image line across the web being viewed contains no defects, only a single record from the twelfth camera with an EOL flag is placed in the FIFOs 346.
The twenty-four-bit records assembled in the FIFOs 346 are transferred over the bus 40 to image processing unit 42 in eight-bit sequential bytes since the image proces_ing unit 42, such as and OCULUS 500 MS board, can - 25 ~ 2V8~ ~7~
only receive data words having aight bits. The control unit 320 in response to absence of an empty signal EF/
from the FIFOs 346 sequentially applies read signals RA/, RB/ and RC/ to the respective FIFOs 346 to sequential apply the outputs of these FIFOs to the bus 40. The control unit also generates a strobe signal on line 350 of the bus 40 during each read signal RA/, RB/
and RC/ to store the record in the unit 42.
Two additional control lines in the bus 330 carry signals to the control unit 320 for causing the gen~ration of SSPUW/ signals on a line (not shown) in each of the buses 332. One of this control signals causes the control unit to send the SSPUW/ signals to all the cameras at the same time; this is needed to enable data to be sent to the system bus 146 during the PHIL, MWL or TINITL signals. The second of these control signals enables a counting and decoding circuit (not shown) for sequentially generating and applying SSPUW/ signals to the buses 332 in the units 322. This counting circuit is stepped to the count of twelve by the LVAL signal from the encoder 44. During each SSPUW/
signal, the signals ALARM1/ - ALARMg/ from the camera channels are applied through a bus 354 and tristate buffers 356 to the bus 330 so that the computer can read the alarm status of the eight channels in each of the cameras. The computer can periodically generate RESTART/ signals and later read the alarm status of the camera channels to monitor the edges of the web. A more precise determination of the edge position can be made by analyzing the defect pixel data generated immediately after a RESTART/ signal and before the edge channels shut down. During the SSPUW/ signal, transmission of defect data records from the corresponding camera are disabled, and generally records corresponding to one image line are lost. Cycling through all twelve cameras substantially insures that the records being collected by each o~ the units 322 generally lose the same number - . , . .
: ~ :
2 ~ 7 ~
of lines so that the r~cords read during one cycle from the units 322 relate to a single image line or adjacent image lines on the web.
The camera alarm signal ALARM in each bus 340 is ORed with the HF/ signal from the corresponding FIFOs 336 by the control unit 320. These ORed alarm signals are also made available for being read over bus 330 by the computer so that the compuker can easily monitor the alarm status. Normally only the extreme edge cameras, cameras one and twelve, will generate alarm signals, and if one of the other cameras generate an alarm, the computer can readily determine the cause of the alarm.
A counter 360 is included in the circuit of Fig. 10 for counting the encoder pulses LVAL. The count is made available to the computer over the bus 330 so that the computer can compare the count of counter 360 with the count made by the image processing unit 42 from records with EOL flags.
In the image processing unit 42, the records being inputted over bus 40 are monitored for defects. EOL
records are used to maintain a line count which indicates the Y-position or coordinate of the image line corresponding to the record. If defect records are uncovered, those with adjacent X-coordinates are assembled and placed in a proper relative address in a video buffer in the processing unit 42 so that the monitor displays the defect. Also the assembled defect data with corresponding X and Y coordinates is transferred to memory common with the computer 46 so that the computer can use the algorithms of International Application No. WO 91/14173 to identi~y the type of defect and make a record which can then be printed in a conventional manner.
Since many modifications, variations, and changes in detail may be made to the above described embodiment~
it is intended that the foregoing description and the - 27 ~ 3 ~ ~ ~
accompanying drawings be interpreted as only illustrative and not in a limiting sense.
:
:
:
~: ' " ',' ;~ : ': : ' :
,`. .. .
-,
with the eight-bit gray value portion o~ output data from the FIF0 130a to the computer interface circuit 70 of Fig. 2. Clock inputs of latches 140, 142 and 144 are connected to respective control lines 148 (PHIL), 150 (MWL) and 152 (TINITL) from the computer interface circuit for storing operating values transferred over the system bus SB from the central computer 46, Fig~ 1, through the interface unit 38. The value ~u is receivPd and applied by latch 140 to the compare circuit 108.
The value ~L iS received and applied by latch circuit 142 to the compare circuit 110. An initial pixel value is received in the latch 144 for flushing prior pixel values in the pipeline and for presenting the initial prior pixel value upon start of the detection of line pixel values. The strobe input 94 is inverted by an inverter 168 to form a clock signal CLK3 which operates various latches in Fig. 6a and operates the counter 126a in Fig. 5.
An encoder pulse input 45 (LV~L) from the~timing circuit 68 of Fig. 2 is high (see also Fig. 7j when data on pixel input 92 is the actual digital value of a pixel in a video line being outputted from the analog-to-digital converter units 64. The input LVAL is delayed by two stages of a latch circuit 182 operated by the strobe 94 to operate four-bit selectors 184a and 184b which have first inputs connected to the output of the latch 144 and second inputs connected to an output of an eight-bit latch 185 receiving the input pixels 92. The outputs of the selectors 184a and 184b are connected to the minuend inputs of the subtracting circuits lOOa and lOOb as well AS to the input of latch 104. The subtrahend inputs of the subtracting circuits lOOa and .
~, . . . . .
!
lOOb are connected to the outputs of the selectors 102a and 102b.
The carry or borrow output 186 of the unit lOOa is a ninth bit delayed by the latch 107 and a stage of latch 136 to enable a NAND gate 188 via inverter 190 when the carry signal 186 is low indicating that the minuend P; is equal to or greater than the subtrahend P~n and to enable gate 192 when the carry signal is high indicating a negative difference value. This gating is necessary since negative differences outputted from subtract circuits lOOa and lOOb are twos complements.
When the gate 188 is enabled, the greater than output of the upper limit comparator 108 is passed through a stage of the latch 136 and the gate 18~ so that the gate 114 passes the result from the upper limit comparator as the defect signal 120. When the gate 192 is enabled, tha less than output of the lower limit comparator llO is passed through the latch 136 and the gate 192 so that the gate 114 passes the result from the lower limit comparator as the defect signal 120.
The use of a prior pixel value as a reference value for a dynamic range of acceptable values in the circuit of Figs. 5 and 6a is based upon a finding that normal pixel values vary very slowly in plastic films being analyzed. However, different testing conditions, such as larger induced noise levels, could cause excessive errors in defect detection. Under such different conditions, errors in defect detection can be reduced by the employment of a pixel averaging circuit (not shown) or a filtering circuit (not shown) such as a finite impulse response (FIR) filter (not shown) or an infinite impulse response (IIR) filter (not shown) in series with latches 104 and 106 so that the prior pixel value presented to the selector 102 is an average or filt~red value derived from one or more prior pixels.
The delayed LVAL signal from latch 182 is further delayed by three staqes of latch 136 and applied by - 15 ~ r~
inverter 198 as the signal STENDEL/ to the CLR input of the counters 125a and 126b. In the absence of incoming pixels, i.e. LVAL being low, the counters 126a and 126b are reset to zero. When the pixels are outputted from the sensor in reverse order as described herein, the outputs of the counters 126a and 126b are to be inverted to correspond to the X-position of the incoming pixels;
due to hardware limitations, this inversion is performed by software in the image processing unit 42 rather than by count-down counters or inverters in the camera. If the incoming pixels were not reversed, such inversion or count-down counter is unnecessary.
The defect signal 120 is applied by an inverter 202 to one input of a NAND gate 204 which enables a NAND
gate 206 to pass the strobe signal 94 to output FCLK.
The output FCLK is applied to one input of a NAND gate 208, Fig.6b, which, when enabled, generates a write signal W/ to operate the FIFOs 130a and 130b, Fig. 5, to store the value of the defect pixel on bus 132 from latch 138 in FPGA 136 and to store the position or X-coordinate on bus 128 from the counters 126a and 126b.
Additionally the defect signal is further delayed by another stage of the latch 136 and applied by an inverter 210 to a second input of NAND gate 204 so that the position and value of the first normal pixel following a series of one or more defect pixels is stored in the FIFOs 130a and 130b. The value and position of this normal pixel are transmitted along with the dPfect pixel data from the cameras to the second stage processing apparatus to enable the further processing of this data.
The thrice delayed signal LVAL from latch 136 is applied by inverter 212 to an output EOLBIT which is applied as one bit input to the FIFO 130b. The twice delayed LVAL signal from latch 136 is inverted by an inverter 214 and applied to one input of a NAMD gate 216 which receives the thrice delayed LVAL signal on its ., ~ .
- 16 - ~ ~83~
other input. The output of the NAND yate 216 is inverted by an inverter 220, delayed by another st~ge in latch 136, inverted by an inverter 222, and applied to a third input of NAND gate 204 to store a high EOLBIT
signal in the FIFO 13Ob along with the position zero and the initial value from latch 144 after the last incoming pixel value of a line section has been tested for a defect value. During the evaluation of pixels in a line section, the EOLBIT signal is low so that the corresponding FIFO stored bit in each defect pixel record is low.
The computer interface has a defect override line 224 (DEFOVRD/ ) connected to a fourth input of the NAND
gate 204 to output all the pixels passing to output 132.
The line 224 is normally high and is rendered low when it is desired to look at pixels or values other than the defect pixels, background pixels, and end of line bits normaIly stored in the FIFOs.
An example of operation of the circuit of Fig. 6a is illustrated in Fig. 7. The upper waveform (CLOCK) represents the clock signal CLK3 distributed to various portions of the circuit. It is noted that Fig. 7 illustrates the relative timing of various circuit portions, but Fig. 7 is not in sufficient detail to accurately refle~t the relative phase shift inherent in the various circuit components of Fig.6a. The values MINUEND, SUBTRAHEND, DIFFERENCE and COUNTER ars set forth in Fig. 7 as decimal numbers, but in the circuit of Fig. 6a these values are presented as parallel eight-bit binary signals. The row DIFFERENCE containsnegative decimal numbers, but in the circuitry of Fig.
6a, negative numbers are the twos compl0ment. The row PIXEL NUMBER in Fig. 7 is inserted for reference purposes to identify the MINUEND image pixels and does not represent any values or signals in Fig. 6a. The COUNTER values represent inverted outputs o~ counters 126a and 126b of Fig. 5.
While LVAL is low, the initial value, one hundred twenty, stored in latch 144 is applied to the minuend inputs of the subtracting circuits lOOa and lOOb. Since this value will have been flushed through the latches 104, 106 and 138, the subtrahend is also equal to one hundred twenty and the difference is zero. When LVAL
goes high indicating that line pixel values are now being applied to latch 185, the selectors 184a and 184b, after a delay, are operated to apply the incoming pixels to the minuend input of the subtracting circuits lOOa and lOOb. The first line image pixel, number 255, has a value of one hundred twenty-one, while pixels, numbers 254-252, have values of one hundred twenty-two. These values do not appear in the subtrahend until three clock periods later so difference values of plus one and plus two are generated. The value ~u is set at plus three and the value -~L iS set at minus three, i.e. llllllO1.
Since plus two is not greater than plus three, the compare high output of the upper limit comparator 108 remains low. The compare low output of lower limit comparator 110 does go high, but the carry signal on line 186 remains low so the gate 192 is disabled to block passage of the compare~low signal to the defect gate 114.
Pixels, numbers 251-248, are defect pixels with values of one hundred sixty-four. Since the difference of plus forty-two is greater than plus three, the compare high output goes high to produce a high defect signal DEFECT. The high defect signal enables the FCLK
signal so that the values of pixels 251-248 along with the pixel X-positions from counters 126a and 126b are stored in the FIFOs 130a and 130b. The high defect signal DEFECT on line 120 also operates the selectors 102a and 102b to prevent the defect values, one hundred sixty-four, being applied to the subtrahend. Rather the subtrahend is held at the last subtrahend value stored in latch 124 before the defect signal, i.e. one hundred , ' -- 18 - ~ 8~ ~L~
twenty-two. In the clock period following the end o~
the defect signal DEFECT, the next normal pixel value and its coordinate are stored in the FIFOs due to the delayed defect signal through inverter 210 to operate gate 204.
Pixels, numbers 9 and 8, have values, one hundred twenty-two, applied to the minuend input which are less than the value, one hundred twenty-three, applied from a prior normal pixel to the subtrahend input. This results in a difference of negative one (1111111~ which is not less than the value -~L (11111101) SO th2 compare low output of the lower limit comparator 110 is low.
The carry signal is high to disable the gate 188 and to prevent the compare high signal from operating the defect gate 114.
Pixels, numbers 7-5, have values, thirty-one, which produce a difference of negative ninety-two (10100100) which is less than negative three (11111101) to produce a high defect signal DEFECT on line 120. This results in the storing of the values of pixels, numbers 7-4, along with their coordinates in the FIFOs.
When LVAL goes low marking the end of line pixel input, the coincidence of inverted twice delayed LVAL
and thrice delayed LVAL operates gate 216 which afker a further delay operates gate 204 to produce a pulse in FCLK to store a further record in the FIFOs 130a and 130b. This record contains the high EOLBIT signal to indicate the end of a line. Also this end of line record contains the initial value from latch 144 which will have been flushed through the pipeline.
The FIFOs 13Oa and 13Ob of Fig. 5 have outputs HF/
and EF/ which, when low, indicate over half full and empty conditions, respectively, of the FIFOs. The half full signal HF/ is applied by inverker 230 in Fig. 6b to the clock input of a flip-flop 232. The inverted output of the flip-flop 232 generates an output ALARM/ which is high in the absence of the half full signal and is ., .
,:
.
applied by OR gate 234 to an inpuk o~ the NAND gate 208.
When the half full signal HF/ goes low, the flip-flop 232 is operated to render the output ALARM/ low and disable the gate ~08. This prevents data of any further defect signals from being written to the FIFOs o~ the channel producing the alarm signal. However, the end o~
line signal EoLsIT is applied to the second input of OR
gate 234 so that end of line records continue to be written in the FIFOs. The size of the FIFOs 130a and 130b, for example 2048 x 9 bit FIFOs, is selected to allow storage of a normal maximum number of defects.
When defect pixels are detected at a rate greater than the rate at which the pixels are read out, the FIFOs over a period of time become half full to shut down storage of further defect pixels.
The alarm signals indicating shutdown of a channel have an additional function, namely that of indicating the edge of the web being examined for defects. The channels viewing image sections outside or straddling the edges of the web detect large numbers of de~ect pixels which quickly produce an alarm signal and shut down the defect detection of those channels~ The alarm signal can be turned off by either a RESTART/ s.ignal or a RST/ signal applied to a respective input of an AND
gate 236 which clears the flip-flop 232. The RST/
signal also resets the latches 182 and 183 in Fig. ~a, operates a flip~flop 238 in Fig. 6b to disable the write signal W/ for the remaining portion of a strobe pulse cycle after termination of the RST/ signal, and is passed of line RS/ to reset the FIFOs 130a and 130b of Fig. 5.
As illustrated in Fig. 2, the computer interface circuit 70, connec~s the defect detection circuits 66 to the cable 34 to the multiplexer and interface unit 38, Fig. 1. The circuit 70, as shown in Fig. 8, includes RS-422 type receivers 240 and 242 and RS-422 type transmitters 244, 246, 248 and 250 connected to lines in .
.
.. ~ .
.
2 ~ 7 ~
the cable 34. The receivers 240 receive control signals RST/, PHIL, TINITL, DEFOVRD/, RESTART/, and MWL
which are passed to the defect detection circuits 66, and receive a control signal SSPUW/ which, along with the signal RST/, is applied to a FPGA 252 controlling the multiplexed reading of the eight defect detectisn channels in circuits 66. The receivers 242 receive eight-bit data signals which are passed to the system bus (RB/SB) 146 when SSPUW/ is low and CAMERAW/ is high, and the transmitters 246 transmit signals on bus RB/SB
146 from FIFO 130a when SSPUW/ is high and CAMERAW/ is low. Transmitters 244 transmit control signals CAMERAW/
and DVALID from the FPGA 252 along with an alarm data signal ALARM Erom NAND gate 254 which has inputs connected to all eight of the alarm outputs of the defect detection circuits. Additionally the transmitters 244 transmit the end of line bit EOL and three bits X1, X2 and X3 which identify which one of the eight data detection channels is being read. The X-coordinate data in the defect records being read are transmitted from bus CB connected to the data outputs of the FIFOs 130b in the eight defect detection circuits when the signal CAMERAW/ is low. When CAMERAW~ is high, the transmitters 248 pass the alarm signals ALARMlj -ALARM8/ from the eight defect detection channels tv the RS-422 lines in cable 34 shared with the outputs of transmitters 250.
In the FPGA 252 as shown in Fig. 9, the outputs Xl, X2 and X3 of a counter are dacoded by conventional decoding circuitry 262 which through NAN~ gates 264 enabled when the signal SSPUW/ is high generates a corresponding one of the signals REN1/ - REN8/.
Referring back to Fig. 6b, the RENX input in each de~ect detection circuit is connected by a jumper (not shown) to a corresponding one of the eight lines RENl/ - REN8/, Fig. 9, to provide for multiplexing the reading of the defect data in the FIFOs 130a and 130b of the eight - 21 - ~ 7~
defect detection circuits. The input RENX/ is applied by inverter 272 to one input of an AND gate 274 which has a second input receiving the master clock signal CLK
from the timing circuit 68, Fig. 2. Third and fourth inputs of the AND yate 274 are connected to the line RST/ and the output of the flip-flop 238 to disable reading during resetting of the circuit. The output of the AND gate 274 is passed throuyh a NAND gate 276 to the line R/ to read records from the FIFOs 130a and 130b onto the buses CB and RB/SB until an EOL bit is produced. The second input of the NAND gate 276 is connected to an output of a flip-flop 278 which is normally high due to the strobe signal applied through inverter 280 to the clock input of the flip-flop 278. A
FIFO empty signal line EF/ from FIFO 130b is connected by inverter 282 to the clear input of a flip-flop 284 which, in the absence of the empty signal, is held clear to prevent its inverted output from operating the clear input of flip-flop 278. However when EF/ goes low, the flip-flop 28~ goes high on the next pulse from NAND gate 276 to clear flip-flop 278 and to disable gate 276 and prevent the reading of empty FIFOs. When EF/ goes high indicating that the FIFOs are not empty, the flip-flop 284 is cleared to remove the clear signal from flip-flop 278 and permit flip-flop 278 to be set high by the next strobe pulse.
The RENX/ signal is also applied by an inverter 286 to an enable input of a tristate buffer 288 which then passes the Rj signal from an inverter 290 to the signal line DVAL which indicates the presence of valid defect data~on the CB and RB/SB buses.
The EOL bit from each FIFO 13Ob is applied to the clock enable input of the counter 260 which is then stepped by the DVAL pulse on its clock input to advance the count in the counter 260. This advances control tv the next REN2/ - REN~/ control line until the last channel has been read. An AND qate 294 has inputs from ~ ~ .
.
.
- 22 ~
all the signals REN1/ - REN8/ to produce the CAMERAW/
signal during readout of the defect data uncoversd in an image line. An inverter 296 applies the output of the AND gate 294 to one input of an A~D gate 298 to enable the gate 298 to pass the signal DVAL through inverters 302 and 304 as the signal DVALID. A third input to the AND gate 298 is received from a NAND gate 306 which has inputs connected to the line REN8/ and to the line ~OL
to disable the gate 298 during each EOL signal except when the circuit 252 i5 reading the last deect detection channel in the camera. Thus the records with high EOL bits from the FIFOs are not recognized by the multiplexer and interface unit 38 as being valid data except for the records with EOL signals from the last camera channel and only one record with an EOL bit high is received as valid data by the multiplexer and interface unit 38 for each image line read out of a camera. If an image line analyzed by the camera contains no pixels of a defect, only a singIe record with an end of line bit high is transmitted from the camera to the multiplexer and interface unit 38.
Each pixel record in the FIFOs 130a and 130b includes on end of line bit (EOL), eight bits of pixel coordinate, and eight bits of pixel intensity. Three additional coordinate bits, identifying the channel in the camera, are added to each data record in transmission to the multiplexer and interface unit 38 to bring the total number of bits to twenty during this transmission. The multiplexer and interface unit 38 adds ~our more coordinate bits identifying the camera from which the data record is read to increase the record bits to twenty-four total bits transmitted from the unit 38 to the image processing unit 42.
As illustrated in Fig. l0, the multiplexer and interface unit 38 contains a central control 320 connected by individual control and data lines to a plurality of substantially similar printed circuit units 322 corresponding to the plurality of ~ameras 30, Fig.
1. Each of the units 322 include RS-422 type transmitters 322 and receivers 3~4 connected to appropriate lines in the corresponding cable 34. The system is under the general control of the computer 46, Fig. 1, through the bus 54 connected to input/output circuit 328. Bus 330 connected to the circuit 328 includes RST/ and RESTART/ lines (not ~hown~ connected to control 320 as well as to each of the units 322; a RST/ signal resets the entire system while a RESTART/
signal clears the alarm flags caused by the HF/ signals from the FIFOs. The bus 330 also includes PHIL, MWL, TINITL and DEFOVRD lines (not shown) along with camera address lines Po-P3 (not shown) connected to the control circuit 320. Buses 332 each including PHIL, MWL, TINITL
and DEFOVRD lines (not shown) are connected from the control circuit 320 to the transmitters in the respective units 322. The bus 330 further contains eight SB data lines (not shown) which are connected to transmitters in block 324 controlled by a CAMERAWX/
signal on line 334 from the control circuit 320. In response to a PHIL, MWL, TINITL or DEFOVRD signal over bus~330 from the computer, the control unit 320 sends the corresponding PHIL, MWL, TINITL or DEFOVRD over the bus 332 to the unit 322 addressed by the signals P0-P3 so as to enable the computer to set the values in latches 140, 142 and 144, Fig. 6a, in the selected cameras to the values SB on the system bus 146.
Each of the units 322 includes three FIFOs 336 to which the twenty incoming data lines in bus 338 from receivers 326 are connected for receiviny and temporarily storing the defect data records from the corresponding camera including an EOL flag bit, eight pixel intensity value bits RB and eleven X-coordinate bits CB. A bus 340 including DVALID, CAMERAWX/ and ALARM lines (not shown) is connected from the receivers 326 to the control circuit 320 which in response to the , . . .
~3~
DVALID and CAMEP~WX/ signals applies write signals to a line W/ to stored each incoming record in the FIFOs 336.
The control circuit also contains circuitry, similar to that of Fig. 6b, connected to a line HF/ from the FIFOs 336 for blocking the write signals, except when a signal is on the line EOL, in response to the FIFOs becoming half full to indicate excessive defects being detected by the corresponding camera. Production of write signals can be restarted by a RESTART/ or a RST/ signal from the computer.
Data outputs of the FIFOs 336 in all the units 322 are connected to a common twenty line bus 344 to inputs of FIFOs 346 which have four additional inputs connected to a bus 348 from the control unit 320 carrying camera address signals. In response to the absence of an EF/
or empty signal from the FIFOs 336, the control unit 320 generates read signals R/ applied to the FIFOs 336 in one selected unit 322 to output defect data records from that selected unit 322 until a record contains an EOL
flag. Simultaneously, the control unit 320 generates write signals on a line W/ to the FIFOs 346~to store the records being read from the selected unit 322 into the FIFOs 346. The control unit 320 contains circuitry similar to Fig. 9 responsive to the EOL signal on bus 344 for suppressing the write signal W/ to FIFOs 346, except when the twelfth unit 322 is being read. Thus records of defects in each image line extending across the view of all twelve cameras are assembled in the FIFOs 246 with an end record having the EOL bit flag from the tweIfth camera. If an image line across the web being viewed contains no defects, only a single record from the twelfth camera with an EOL flag is placed in the FIFOs 346.
The twenty-four-bit records assembled in the FIFOs 346 are transferred over the bus 40 to image processing unit 42 in eight-bit sequential bytes since the image proces_ing unit 42, such as and OCULUS 500 MS board, can - 25 ~ 2V8~ ~7~
only receive data words having aight bits. The control unit 320 in response to absence of an empty signal EF/
from the FIFOs 346 sequentially applies read signals RA/, RB/ and RC/ to the respective FIFOs 346 to sequential apply the outputs of these FIFOs to the bus 40. The control unit also generates a strobe signal on line 350 of the bus 40 during each read signal RA/, RB/
and RC/ to store the record in the unit 42.
Two additional control lines in the bus 330 carry signals to the control unit 320 for causing the gen~ration of SSPUW/ signals on a line (not shown) in each of the buses 332. One of this control signals causes the control unit to send the SSPUW/ signals to all the cameras at the same time; this is needed to enable data to be sent to the system bus 146 during the PHIL, MWL or TINITL signals. The second of these control signals enables a counting and decoding circuit (not shown) for sequentially generating and applying SSPUW/ signals to the buses 332 in the units 322. This counting circuit is stepped to the count of twelve by the LVAL signal from the encoder 44. During each SSPUW/
signal, the signals ALARM1/ - ALARMg/ from the camera channels are applied through a bus 354 and tristate buffers 356 to the bus 330 so that the computer can read the alarm status of the eight channels in each of the cameras. The computer can periodically generate RESTART/ signals and later read the alarm status of the camera channels to monitor the edges of the web. A more precise determination of the edge position can be made by analyzing the defect pixel data generated immediately after a RESTART/ signal and before the edge channels shut down. During the SSPUW/ signal, transmission of defect data records from the corresponding camera are disabled, and generally records corresponding to one image line are lost. Cycling through all twelve cameras substantially insures that the records being collected by each o~ the units 322 generally lose the same number - . , . .
: ~ :
2 ~ 7 ~
of lines so that the r~cords read during one cycle from the units 322 relate to a single image line or adjacent image lines on the web.
The camera alarm signal ALARM in each bus 340 is ORed with the HF/ signal from the corresponding FIFOs 336 by the control unit 320. These ORed alarm signals are also made available for being read over bus 330 by the computer so that the compuker can easily monitor the alarm status. Normally only the extreme edge cameras, cameras one and twelve, will generate alarm signals, and if one of the other cameras generate an alarm, the computer can readily determine the cause of the alarm.
A counter 360 is included in the circuit of Fig. 10 for counting the encoder pulses LVAL. The count is made available to the computer over the bus 330 so that the computer can compare the count of counter 360 with the count made by the image processing unit 42 from records with EOL flags.
In the image processing unit 42, the records being inputted over bus 40 are monitored for defects. EOL
records are used to maintain a line count which indicates the Y-position or coordinate of the image line corresponding to the record. If defect records are uncovered, those with adjacent X-coordinates are assembled and placed in a proper relative address in a video buffer in the processing unit 42 so that the monitor displays the defect. Also the assembled defect data with corresponding X and Y coordinates is transferred to memory common with the computer 46 so that the computer can use the algorithms of International Application No. WO 91/14173 to identi~y the type of defect and make a record which can then be printed in a conventional manner.
Since many modifications, variations, and changes in detail may be made to the above described embodiment~
it is intended that the foregoing description and the - 27 ~ 3 ~ ~ ~
accompanying drawings be interpreted as only illustrative and not in a limiting sense.
:
:
:
~: ' " ',' ;~ : ': : ' :
,`. .. .
-,
Claims (32)
1. A defect detection system comprising a camera housing, an array sensor mounted in the camera housing, a lens mounted on the housing for projecting an image of an object under test onto the array sensor, means for operating the array sensor to generate a video signal representing at least a portion of the object, means within the housing for detecting pixels in the video signal of a defect in the object, and camera output means in the housing for transmitting the detected defect pixels apart from normal pixels in the video signal whereby the quantity of pixel values transmitted by the camera output means is less than the quantity of pixel values in the video signal.
2. A defect detection system as claimed in claim 1 wherein the camera output means includes means for temporarily storing the detected defect pixels prior to transmitting the detected defect pixels.
3. A defect detection system as claimed in claim 2 wherein the temporary storing means includes a first-in first-out memory.
4. A defect detection system as claimed in claim 1 wherein means for detecting defect pixels in the video signal includes means for determining if each pixel in the video signal is within an acceptable range of pixel values.
5. A defect detection system as claimed in claim 4 including means responsive to a change in pixels within the acceptable range of pixel values for changing the acceptable range of pixel values so that the acceptable range of pixels values follows gradual changes in the pixels of the video signal.
6. A defect detection system as claimed in claim 4 wherein the means for determining if each pixel is within the acceptable range includes a subtraction circuit receiving a value of each present pixel and a prior pixel value derived from at least one prior pixel for producing a difference between the present pixel value and the prior pixel value, and a comparison circuit receiving the difference from the subtraction circuit for producing a defect signal if the difference exceeds a predetermined value.
7. A defect detection system as claimed in claim 6 including means for temporarily storing the prior pixel value derived from at least one prior pixel and presenting the stored prior pixel value to the subtraction circuit, means responsive to the absence of the defect signal for updating the stored prior pixel value to a succeeding prior pixel value derived from at least one prior pixel value, and means responsive to the defect signal for maintaining the stored prior pixel value at its present value.
8. A defect detection system as claimed in claim 7 wherein the means for temporarily storing the prior pixel value derived from at least one prior pixel includes means for storing a value of a prior pixel.
9. A defect detection system comprising an array sensor including operating means for generating a video signal of an image imposed on the array sensor, lens means for projecting an image of an object under test onto the array sensor, means for temporarily storing a prior pixel value derived from at least one pixel value prior to a present pixel value in the video signal, a subtraction circuit receiving the present pixel value and the stored prior pixel value for producing a difference between the present pixel value and the stored prior pixel value, a comparison circuit receiving the difference from the subtraction circuit for producing a defect signal if the difference exceeds a predetermined value, and means responsive to the defect signal for storing defect pixel values apart from normal pixel values.
10. A defect detection system as claimed in claim 9 further including means responsive only to the absence of a defect signal for operating the prior pixel temporary storing means to store a succeeding prior pixel value derived from at least one pixel value prior to a present pixel value so that the stored prior pixel value remains constant during a defect signal.
11. A defect detection system as claimed in claim 9 wherein the means for storing the defect pixel values includes means for storing coordinates of defect pixels.
12. A defect detection system as claimed in claim 9 wherein the means for storing the defect pixel values includes a first-in first-out memory.
13. A defect detection system as claimed in claim 9 wherein the subtraction circuit includes an output producing a signal indicating positive and negative differences; and the comparison circuit includes an upper limit comparator for comparing the difference with a predetermined upper limit, a lower limit comparator for comparing the difference with a lower limit, and logic means connected to a greater-than-output of the upper limit comparator, a less-than-output of the lower limit comparator, and the positive and negative indicating output for passing the greater-than-output as the defect signal when the difference is positive and for passing the less-than-output as the defect signal when the difference is negative.
14. A defect detection system comprising a TDI CCD two-dimensional array sensor having a plurality of outputs each for generating a video signal of a corresponding section of an image of a moving object, a plurality of defect detecting circuits connected to the respective plurality of outputs of the array sensor, a plurality of temporary memories for storing defect pixel values detected by the respective plurality of defect detecting circuits, and output means for reading the temporary memories to transfer the defect pixel values.
15. A defect detection system as claimed in claim 14 wherein the output means sequentially reads the temporary memories to sequentially transfer the defect pixel values.
16. A defect detection system as claimed in claim 14 wherein each temporary memory includes means for storing a coordinate for each detected defect pixel value, and the output means reads and transfers the coordinates along with the defect pixel values.
17. A defect detection system as claimed in claim 16 wherein the output means includes means for expanding the coordinates prior to transfer to include the position of the image section containing the detected defect pixel.
18. A defect detection system as claimed in claim 14 wherein each temporary memory includes means for storing an end of line signal; and the output means includes means for sequentially selecting the plurality of temporary memories to read the memories, and means responsive to an end of line signal from a temporary memory for selecting the next temporary memory.
19. A defect detection system as claimed in claim 18 wherein the sequential selecting means includes a counter for counting the end of line signals.
20. A defect detection system as claimed in claim 19 wherein each temporary memory includes means for storing a coordinate for each detected defect pixel value; and the output means includes means for reading the coordinates along with the defect pixel values, means for incorporating the output of the counter into the coordinates, and means for transferring the coordinates with the incorporated counter output along with the detected defect pixel values.
21. A defect detection system as claimed in claim 14 including a camera housing enclosing the TDI CCD two-dimensional array sensor, the plurality of defect detection circuits, the plurality of temporary memories, and the output means.
22. A defect detection system comprising an array sensor for generating successive pixel values of an image portion of a moving object, a defect detecting circuit connected to the array sensor for detecting defective pixel values in the pixel values of the image portion, a temporary memory, means responsive to the defect detecting circuit detecting defective pixel values for storing the defective pixel values in the temporary memory, means for reading out the stored defective pixel values from the temporary memory at a rate which is less than the rate at which the array sensor generates the pixel; values, and means for determining the presence of more than a predetermined number of unread stored defective pixel values in the temporary memory to disable the defective pixel value storing means.
23. A defect detecting system as claimed in claim 22 wherein the array sensor generates successive pixel values of each of a plurality of image lines of the image portion, the system includes means for storing an end of line flag in the temporary memory for each image line sensed by the sensor irrespective of the number of unread stored defective pixel values, and the reading out means reads out each stored end of line flag after reading out any stored defective pixels corresponding to each end of line flag.
24. A defect detecting system as claimed in claim 23 including means responsive to the end of line flag reading out means for determining a coordinate of the object.
25. A defect detection system as claimed in claim 22 wherein the means for determining the presence of more than a predetermined number of unread stored defective pixel values includes means for maintaining the defective pixel storing means disabled after the reading out means reduces the number of unread stored defective pixel values to less than the predetermined number.
26. A defect detecting system as claimed in claim 22 wherein the temporary memory is a first-in, first-out memory having a half-full flag, and the means for determining the presence of more than a predetermined number of unread stored defective pixel values responds to the half-full flag to determine the presence of more than the predetermined number.
27. A defect detecting system as claimed in claim 22 wherein the array sensor is a TDI CCD two-dimensional array sensor having a plurality of outputs each for generating successive pixel values of a corresponding section of an image of a moving object; and the system includes a plurality of defect detecting circuits connected to the respective plurality of outputs of the array sensor for detecting defective pixel values in the pixel values from the respective outputs, a plurality of temporary memories corresponding to the plurality of detecting means, a plurality of means responsive to the respective defect detecting circuits for storing the defective pixel values in the respective temporary memories, means for reading out stored defective pixel values from the plurality of memories one memory at a time, and a plurality of means for determining the presence of more than a predetermined number of unread stored defective pixel values in the respective memories to disable the respective defective pixel value storing means.
28. A defect detection system as claimed in claim 27 wherein each output of the array sensor generates successive pixel values of each of a plurality of image lines of the corresponding image section; and the system includes a plurality of means for storing end of line flags in the respective memories for each image line of the respective sections irrespective of the number of unread stored defective pixel values, and the reading out means reads out each stored end of line flag after reading out any stored defective pixels of an image line corresponding to each end of line flag and sequences to the reading out of the next memory in response to each end of line flag.
29. A defect detection system as claimed in claim 28 wherein the reading out means includes means for discarding the end of line flags from all but one designated memory of the plurality of memories so that all defect pixels on a common line through the image sections are sequentially outputted with a single end of line flag.
30. A defect detection system as claimed in claim 27 including means for determining an edge of the object by sensing which of the plurality of means for determining the presence of more than a predetermined number of unread stored defective pixel values in the respective memories have operated.
31. A defect detection system as claimed in claim 27 wherein each of the plurality of means for determining the presence of more than a predetermined number of unread stored defective pixel values includes means for maintaining the respective defective pixel storing means disabled after the reading out means reduces the number of unread stored defective pixel values in the respective memory to less than the predetermined number.
32. A defect detecting system as claimed in claim 27 wherein each of the temporary memories is a first-in, first-out memory having a half-full flag, and the plurality of means for determining the presence of more than a predetermined number of unread stored defective pixel values responds to the respective half-full flags to determine the presence of more than the predetermined number of unread stored defective pixel values in the respective memories.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US07/794,861 US5440648A (en) | 1991-11-19 | 1991-11-19 | High speed defect detection apparatus having defect detection circuits mounted in the camera housing |
US794,861 | 1991-11-19 |
Publications (1)
Publication Number | Publication Date |
---|---|
CA2083074A1 true CA2083074A1 (en) | 1993-05-20 |
Family
ID=25163904
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CA002083074A Abandoned CA2083074A1 (en) | 1991-11-19 | 1992-11-17 | Defect detection |
Country Status (5)
Country | Link |
---|---|
US (1) | US5440648A (en) |
EP (1) | EP0543629B1 (en) |
JP (1) | JPH05304674A (en) |
CA (1) | CA2083074A1 (en) |
DE (1) | DE69232476T2 (en) |
Families Citing this family (92)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5668887A (en) * | 1992-05-29 | 1997-09-16 | Eastman Kodak Company | Coating density analyzer and method using non-synchronous TDI camera |
EP0572336B1 (en) * | 1992-05-29 | 2001-03-14 | Eastman Kodak Company | Coating density analyzer and method using image processing |
US5859923A (en) * | 1992-12-29 | 1999-01-12 | Cognex Corporation | Mark quality inspection apparatus and method |
JP3312974B2 (en) * | 1993-11-24 | 2002-08-12 | 吉郎 山田 | Imaging device |
US5600574A (en) * | 1994-05-13 | 1997-02-04 | Minnesota Mining And Manufacturing Company | Automated image quality control |
USH1616H (en) * | 1994-05-31 | 1996-12-03 | Minnesota Mining And Manufacturing Company | Web inspection system having enhanced video signal preprocessing |
US5878153A (en) * | 1994-06-24 | 1999-03-02 | Ford Global Technologies, Inc. | Method for monitoring coating adhesion propensity based on surface morphology |
US5880772A (en) * | 1994-10-11 | 1999-03-09 | Daimlerchrysler Corporation | Machine vision image data acquisition system |
KR0181286B1 (en) * | 1995-03-28 | 1999-05-01 | 김광호 | Defect compensating circuit for ccd camera |
US5642198A (en) * | 1995-04-03 | 1997-06-24 | Long; William R. | Method of inspecting moving material |
PT741290E (en) * | 1995-05-04 | 2000-06-30 | Mahlo Gmbh & Co Kg | METHOD AND APPARATUS FOR DETERMINING THE MOTION DIALING ANGLES IN MOTION OR SIMILAR APPLICATIONS |
ATE190722T1 (en) * | 1995-05-10 | 2000-04-15 | Mahlo Gmbh & Co Kg | METHOD AND DEVICE FOR DETECTING DEFECTS IN TEXTILE WEAVES AND THE LIKE |
AR002082A1 (en) * | 1995-05-25 | 1998-01-07 | Gillette Co | METHOD TO PRODUCE A CUTTING EDGE IN A METAL STRIP AND THE APPLIANCE TO PERFORM IT |
HU224946B1 (en) * | 1995-09-22 | 2006-04-28 | Owens Brockway Glass Container | System for checking containers by a field programmable gate array logic |
WO1997019217A2 (en) * | 1995-11-21 | 1997-05-29 | Loughborough University Innovations Limited | Control methods and apparatus |
US5696591A (en) * | 1996-01-05 | 1997-12-09 | Eastman Kodak Company | Apparatus and method for detecting longitudinally oriented flaws in a moving web |
GB2310503A (en) * | 1996-02-26 | 1997-08-27 | Robert Joseph Hunter | A rectangular co-ordinate plotter |
FR2746330B1 (en) * | 1996-03-22 | 1998-04-17 | Alcatel Postal Automation Syst | IMAGE ACQUISITION SYSTEM FOR PACKET SORTING |
DE19624905A1 (en) * | 1996-06-21 | 1998-01-08 | L & P Elektroautomatisations G | Piece goods quality control |
ES2117575B1 (en) * | 1996-08-05 | 1999-04-01 | Chep Pooling Systems B V | PROCEDURE FOR THE AUTOMATIC RECOGNITION AND CLASSIFICATION OF DEFECTS IN PALLETS OR SIMILAR ELEMENTS, AND THE CORRESPONDING SYSTEM. |
MX9603554A (en) * | 1996-08-21 | 1998-04-30 | Servicios Condumex Sa | Counting sensing device and surface cracks and bubbles dimensioning on copper bars during continuous melt process. |
US5838840A (en) * | 1996-08-29 | 1998-11-17 | Bst/Pro Mark | Inspection device using a field mode video camera with interpolation to replace missing pixels |
US6381355B1 (en) * | 1997-08-19 | 2002-04-30 | The Hong Kong University Of Science And Technology | Inspection method for comparison of articles |
WO1999010833A1 (en) * | 1997-08-27 | 1999-03-04 | Datacube, Inc. | Web inspection system for analysis of moving webs |
WO1999013322A1 (en) * | 1997-09-08 | 1999-03-18 | Photon Dynamics, Inc. | Color filter inspection method and apparatus |
US6236429B1 (en) * | 1998-01-23 | 2001-05-22 | Webview, Inc. | Visualization system and method for a web inspection assembly |
AT406528B (en) * | 1998-05-05 | 2000-06-26 | Oesterr Forsch Seibersdorf | METHOD AND DEVICE FOR DETECTING, IN PARTICULAR FOR VISUALIZING, ERRORS ON THE SURFACE OF OBJECTS |
US6324298B1 (en) * | 1998-07-15 | 2001-11-27 | August Technology Corp. | Automated wafer defect inspection system and a process of performing such inspection |
US6593961B1 (en) * | 1998-10-30 | 2003-07-15 | Agilent Technologies, Inc. | Test efficient method of classifying image quality of an optical sensor using three categories of pixels |
FI110343B (en) * | 1999-02-25 | 2002-12-31 | Honeywell Oy | Monitoring system for a paper machine's breakage |
US6330343B1 (en) * | 1999-02-26 | 2001-12-11 | Conoco Inc. | Method for measuring coke quality by digital quantification of high intensity reflectivity |
US6954545B2 (en) | 1999-02-26 | 2005-10-11 | Conocophillips Company | Use of a scanner to determine the optical density of calcined coke as a measure of coke quality |
US6259526B1 (en) * | 1999-03-01 | 2001-07-10 | Sara Lee Corporation | Operator-free fabric web inspection system |
JP3454753B2 (en) * | 1999-06-02 | 2003-10-06 | 株式会社キリンテクノシステム | Automatic inspection device by image processing |
DE19953741C1 (en) * | 1999-11-09 | 2001-10-25 | Krones Ag | Device and method for optical inspection |
US6934028B2 (en) * | 2000-01-20 | 2005-08-23 | Webview, Inc. | Certification and verification management system and method for a web inspection apparatus |
JP3879359B2 (en) * | 2000-03-17 | 2007-02-14 | オムロン株式会社 | Image recording device |
US6831998B1 (en) * | 2000-06-22 | 2004-12-14 | Hitachi, Ltd. | Inspection system for circuit patterns and a method thereof |
JP2002024802A (en) * | 2000-07-04 | 2002-01-25 | Nippon Sheet Glass Co Ltd | Image processing system |
US7113627B1 (en) * | 2000-08-09 | 2006-09-26 | Eastman Kodak Company | Location of extended linear defects |
KR100345896B1 (en) * | 2000-11-20 | 2002-07-27 | 삼성전자 주식회사 | Cctv system |
US7052483B2 (en) * | 2000-12-19 | 2006-05-30 | Animas Corporation | Transcutaneous inserter for low-profile infusion sets |
US6750466B2 (en) * | 2001-02-09 | 2004-06-15 | Wintriss Engineering Corporation | Web inspection system |
US6950547B2 (en) * | 2001-02-12 | 2005-09-27 | 3M Innovative Properties Company | Web inspection method and device |
JP4485087B2 (en) * | 2001-03-01 | 2010-06-16 | 株式会社半導体エネルギー研究所 | Operation method of semiconductor device |
EP1249415A3 (en) * | 2001-04-14 | 2004-02-04 | NexPress Solutions LLC | Method and device for measuring positions of continuously moving sheets |
SE0101374L (en) | 2001-04-19 | 2002-10-20 | Svante Bjoerk Ab | Optical inspection method and apparatus |
US7009163B2 (en) * | 2001-06-22 | 2006-03-07 | Orbotech Ltd. | High-sensitivity optical scanning using memory integration |
GB2384852A (en) * | 2001-09-03 | 2003-08-06 | Millennium Venture Holdings Lt | Workpiece inspection apparatus |
US6904330B2 (en) | 2002-08-07 | 2005-06-07 | Kimberly-Clark Worldwide, Inc. | Manufacturing information and troubleshooting system and method |
US7130709B2 (en) * | 2002-08-07 | 2006-10-31 | Kimberly-Clark Worldwide, Inc. | Manufacturing information and alarming system and method |
JP4185789B2 (en) * | 2003-03-12 | 2008-11-26 | 株式会社日立ハイテクノロジーズ | Pattern inspection method and apparatus |
DE10326033B4 (en) * | 2003-06-10 | 2005-12-22 | Hema Electronic Gmbh | Method for adaptive error detection on an inhomogeneous surface |
CA2438951A1 (en) * | 2003-08-29 | 2005-02-28 | Bob Richards | Feeder system and method |
BRPI0418285A (en) * | 2003-12-31 | 2007-05-02 | 3M Innovative Properties Co | method and system for automated continuous sheet inspection, conversion control system, and computer readable media |
WO2005065367A2 (en) * | 2003-12-31 | 2005-07-21 | 3M Innovative Properties Company | Maximation of yield for web-based articles |
US7382457B2 (en) * | 2004-01-22 | 2008-06-03 | Wintriss Engineering Corporation | Illumination system for material inspection |
US7623699B2 (en) * | 2004-04-19 | 2009-11-24 | 3M Innovative Properties Company | Apparatus and method for the automated marking of defects on webs of material |
US7868903B2 (en) | 2004-10-14 | 2011-01-11 | Daktronics, Inc. | Flexible pixel element fabrication and sealing method |
US8344410B2 (en) | 2004-10-14 | 2013-01-01 | Daktronics, Inc. | Flexible pixel element and signal distribution means |
US7893948B1 (en) | 2004-10-14 | 2011-02-22 | Daktronics, Inc. | Flexible pixel hardware and method |
US8001455B2 (en) * | 2004-10-14 | 2011-08-16 | Daktronics, Inc. | Translation table |
US7952633B2 (en) * | 2004-11-18 | 2011-05-31 | Kla-Tencor Technologies Corporation | Apparatus for continuous clocking of TDI sensors |
US20060226380A1 (en) * | 2005-04-11 | 2006-10-12 | Meinan Machinery Works, Inc. | Method of inspecting a broad article |
US7599516B2 (en) | 2005-08-23 | 2009-10-06 | Illinois Tool Works Inc. | Machine vision counting system apparatus and method |
US20070189597A1 (en) * | 2005-08-23 | 2007-08-16 | Limer Daniel J | Machine vision counting system apparatus and method |
FI119708B (en) * | 2006-02-01 | 2009-02-13 | Viconsys Oy | Device for checking a path |
EP1867979B1 (en) * | 2006-06-13 | 2009-03-11 | ABB Oy | Method and apparatus for recognizing repeating patterns |
EP1868366A1 (en) * | 2006-06-16 | 2007-12-19 | THOMSON Licensing | Method for controlling a TDI-CCD image sensor |
BE1017422A3 (en) * | 2006-12-08 | 2008-09-02 | Visys Nv | Product e.g. raisins, sorting method, involves capturing light reflected by products in product stream, and automatically separating products from product stream based on captured light |
JP4417400B2 (en) * | 2007-04-16 | 2010-02-17 | アンリツ株式会社 | Solder inspection line centralized management system and management device used therefor |
US8175739B2 (en) * | 2007-07-26 | 2012-05-08 | 3M Innovative Properties Company | Multi-unit process spatial synchronization |
US7542821B2 (en) * | 2007-07-26 | 2009-06-02 | 3M Innovative Properties Company | Multi-unit process spatial synchronization of image inspection systems |
US20090028417A1 (en) * | 2007-07-26 | 2009-01-29 | 3M Innovative Properties Company | Fiducial marking for multi-unit process spatial synchronization |
JP5100419B2 (en) * | 2008-01-30 | 2012-12-19 | オリンパス株式会社 | Inspection system |
JP4537467B2 (en) * | 2008-03-18 | 2010-09-01 | アドバンスド・マスク・インスペクション・テクノロジー株式会社 | Sample inspection apparatus and sample inspection method |
US8537353B2 (en) * | 2008-05-05 | 2013-09-17 | Agency For Science, Technology And Research | Sensor chip for biological and chemical sensing |
US7797133B2 (en) * | 2008-09-10 | 2010-09-14 | 3M Innovative Properties Company | Multi-roller registered repeat defect detection of a web process line |
US8624971B2 (en) * | 2009-01-23 | 2014-01-07 | Kla-Tencor Corporation | TDI sensor modules with localized driving and signal processing circuitry for high speed inspection |
US8654333B2 (en) * | 2010-03-30 | 2014-02-18 | Fujifilm Corporation | Surface inspection apparatus and method |
TW201205180A (en) * | 2010-07-27 | 2012-02-01 | Hon Hai Prec Ind Co Ltd | Camera |
CN102156990B (en) * | 2011-04-02 | 2013-12-11 | 北京理工大学 | Automatic identification method for blur parameters of TDI-CCD aerial remote sensing image |
US8922641B2 (en) | 2011-06-29 | 2014-12-30 | The Procter & Gamble Company | System and method for inspecting components of hygienic articles |
US9213227B2 (en) * | 2011-08-18 | 2015-12-15 | Nikon Corporation | Custom color or polarization sensitive CCD for separating multiple signals in autofocus projection system |
US8682047B2 (en) | 2011-12-05 | 2014-03-25 | Illinois Tool Works Inc. | Method and apparatus for machine vision counting and annotation |
US8861816B2 (en) | 2011-12-05 | 2014-10-14 | Illinois Tool Works Inc. | Method and apparatus for prescription medication verification |
US9860466B2 (en) | 2015-05-14 | 2018-01-02 | Kla-Tencor Corporation | Sensor with electrically controllable aperture for inspection and metrology systems |
US10620052B2 (en) * | 2015-08-10 | 2020-04-14 | United States Gypsum Company | System and method for manufacturing cementitious boards with on-line void detection |
US10778925B2 (en) | 2016-04-06 | 2020-09-15 | Kla-Tencor Corporation | Multiple column per channel CCD sensor architecture for inspection and metrology |
US10313622B2 (en) | 2016-04-06 | 2019-06-04 | Kla-Tencor Corporation | Dual-column-parallel CCD sensor and inspection systems using a sensor |
FI129412B (en) * | 2018-04-13 | 2022-01-31 | Maillefer Sa | An arrangement and a method for surface defect detection of a cable |
CN113012128B (en) * | 2021-03-18 | 2022-04-15 | 长鑫存储技术有限公司 | Defect characterization method and device |
Family Cites Families (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US3920970A (en) * | 1974-04-30 | 1975-11-18 | Intec Corp | Laser scanner flaw detection system using baseline follower signal processing |
JPS51113783A (en) * | 1975-03-31 | 1976-10-07 | Asahi Chem Ind Co Ltd | Defects detector of nonwoven fabric |
US4046536A (en) * | 1976-08-13 | 1977-09-06 | Western Electric Company, Inc. | Monitoring and control of optical fiber diameters |
JPS5546172A (en) * | 1978-09-29 | 1980-03-31 | Kirin Brewery Co Ltd | Detector for foreign material |
JPS5585028A (en) * | 1978-12-22 | 1980-06-26 | Hitachi Ltd | Mark detecting signal amplifier |
US4433346A (en) * | 1982-03-31 | 1984-02-21 | Xerox Corporation | Raster input scanner |
DE3267548D1 (en) * | 1982-05-28 | 1986-01-02 | Ibm Deutschland | Process and device for an automatic optical inspection |
US4546444A (en) * | 1983-03-15 | 1985-10-08 | E. I. Du Pont De Nemours And Company | Data compression interface having parallel memory architecture |
DE3578768D1 (en) * | 1985-03-14 | 1990-08-23 | Toppan Printing Co Ltd | DEVICE FOR CHECKING PRINTING. |
DE3636192C1 (en) * | 1986-10-24 | 1988-02-18 | Erhardt & Leimer Gmbh | Device for capturing the image of surface areas of running webs of goods |
US4950911A (en) * | 1986-12-10 | 1990-08-21 | Process Automation Business, Inc. | Apparatus and method for inspecting sheet material |
US4752897A (en) * | 1987-05-01 | 1988-06-21 | Eastman Kodak Co. | System for monitoring and analysis of a continuous process |
US4764681A (en) * | 1987-06-04 | 1988-08-16 | Owens-Illinois Televison Products Inc. | Method of and apparatus for electrooptical inspection of articles |
GB8822183D0 (en) * | 1988-09-21 | 1988-10-26 | Radix Systems Ltd | Method & apparatus for processing line scan optical images |
US5040057A (en) * | 1990-08-13 | 1991-08-13 | Picker International, Inc. | Multi-mode TDI/raster-scan television camera system |
US4949172A (en) * | 1988-09-26 | 1990-08-14 | Picker International, Inc. | Dual-mode TDI/raster-scan television camera system |
JP3151449B2 (en) * | 1990-03-13 | 2001-04-03 | マーコニ・メディカル・システムズ・インコーポレイテッド | Web inspection system |
EP0455898A1 (en) * | 1990-05-09 | 1991-11-13 | Robert Bishop | Image scanning inspection system |
FI85308C (en) * | 1990-06-07 | 1992-03-25 | Rautaruukki Oy | Method and apparatus for optical examination of sheet and web-shaped products |
US5184217A (en) * | 1990-08-02 | 1993-02-02 | Doering John W | System for automatically inspecting a flat sheet part |
US5157486A (en) * | 1990-09-21 | 1992-10-20 | Fmc Corporation | High resolution camera sensor having a linear pixel array |
US5132791A (en) * | 1990-09-25 | 1992-07-21 | Ball Corporation | Optical sheet inspection system |
-
1991
- 1991-11-19 US US07/794,861 patent/US5440648A/en not_active Expired - Fee Related
-
1992
- 1992-11-17 CA CA002083074A patent/CA2083074A1/en not_active Abandoned
- 1992-11-18 EP EP92310506A patent/EP0543629B1/en not_active Expired - Lifetime
- 1992-11-18 DE DE69232476T patent/DE69232476T2/en not_active Expired - Fee Related
- 1992-11-19 JP JP4310307A patent/JPH05304674A/en active Pending
Also Published As
Publication number | Publication date |
---|---|
US5440648A (en) | 1995-08-08 |
DE69232476T2 (en) | 2002-10-31 |
EP0543629B1 (en) | 2002-03-13 |
EP0543629A1 (en) | 1993-05-26 |
DE69232476D1 (en) | 2002-04-18 |
JPH05304674A (en) | 1993-11-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US5440648A (en) | High speed defect detection apparatus having defect detection circuits mounted in the camera housing | |
US5812704A (en) | Method and apparatus for image overlap processing | |
US5282268A (en) | Video image storage system | |
US4549205A (en) | Ampoule inspecting method | |
EP0094824B1 (en) | Image processing device for continuously extracting features of small regions of an image | |
US5046120A (en) | Method of and apparatus for increasing the processing speed in the scanning inspection of circuit boards and other objects | |
GB2224831A (en) | System for processing line scan video image signal information | |
US4516263A (en) | Spatially integral, video signal processor | |
JP3349037B2 (en) | Method and apparatus for capturing moving object image | |
US20020149684A1 (en) | Method and apparatus for qualification of image detectors | |
JPH03195922A (en) | Detecting apparatus for degree of congestion | |
EP0380090B1 (en) | Image processing system | |
US4092670A (en) | Video processor for solid state star tracker | |
US20060170954A1 (en) | Method and system for generating synchronous multidimensional data streams from a one -dimensional data stream | |
Elphinstone et al. | RAPAC: a high-speed image-processing system | |
JP2725674B2 (en) | Defect judgment device | |
JPS6352267A (en) | State monitoring device | |
JP3574883B2 (en) | Pupil measurement device | |
JPH09284753A (en) | Method and device for controlling image pickup of supervisory camera | |
JP2659783B2 (en) | Image processing device | |
JPS63245084A (en) | Interlace picture data conversion system | |
JP3100141B2 (en) | Image processing method and image processing apparatus | |
JP2836129B2 (en) | Frame synchronization method in image processing | |
JPS59132292A (en) | Picture device | |
Roberts et al. | Programmable CCD camera equipped with user-configurable video rate digital video processing for use in industrial inspection |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
EEER | Examination request | ||
FZDE | Discontinued | ||
FZDE | Discontinued |
Effective date: 20040805 |