|Publication number||USH681 H|
|Application number||US 07/058,515|
|Publication date||Sep 5, 1989|
|Filing date||Jun 5, 1987|
|Priority date||Jun 5, 1987|
|Publication number||058515, 07058515, US H681 H, US H681H, US-H-H681, USH681 H, USH681H|
|Inventors||William E. Weideman|
|Export Citation||BiBTeX, EndNote, RefMan|
|Referenced by (14), Classifications (8), Legal Events (1)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This invention relates to optical character recognition systems, and more particularly to a dot matrix print detector to assist in the character recognition process.
Document processing systems optically read information from documents. Numerous optical character recognition techniques are utilized for distinguishing between different print fonts and for identifying the individual characters themselves. As used herein, the word "characters" shall include alphanumeric characters as well as bar code data. Characters may be printed in a continuous format typically referred to as contiguous print or in a dot matrix print format in which rows and columns of dots are utilized to create a character.
Since character recognition involves numerous parameters associated with a character, whether a character is contiguous or represented by dot matrix print is an important piece of information in recognizing the character. Various template and feature recognition algorithms are utilized for optical character recognition techniques. Different masks for template recognition are utilized to recognize dot matrix print since feature recognition algorithms do not produce the desired results in recognizing dot matrix print characters. Therefore, if the presence of a dot matrix print character were known, algorithms that are specifically directed to the recognition of dot matrix print characters could be utilized instead of methods for recognizing contiguous characters. Further, knowledge of the presence or absence of a dot matrix print character would be useful in character segmentation and contextual post-processing of the recognition results.
A need has thus arisen for a dot matrix print detector for identifying the presence or absence of a dot matrix print character to assist in the character recognition process.
In accordance with the present invention, a dot matrix print detector for use with an optical character recognition system is provided. The detector includes structure for scanning a document containing the characters and for generating pixel signals representing pixel values for portions of the characters appearing on the document. A filter is provided for receiving the pixel signals and for generating a filtered signal representing energy levels of pixels associated with edges of the characters. A summing circuit is provided for determining the energy level of the filtered signals for a predetermined area on the document and for generating a first energy signal. A second summing circuit is provided for determining the energy level of the pixel signals generated by the scanning structure for the predetermined area and for generating a second energy signal. A comparator compares the first and second energy signals for generating a signal indicating the presence of dot matrix print in a character on the document in the predetermined area.
For a complete understanding of the present invention and for further advantages thereof, reference is now made to the following Detailed Description taken in conjunction with the accompanying Drawings in which:
FIG. 1 is an enlarged illustration of a contiguous print character;
FIG. 2 is an enlarged illustration of a dot matrix print character; and
FIG. 3 is a block diagram of the present dot matrix print detector.
Referring simultaneously to FIGS. 1 and 2, FIG. 1 illustrates a contiguous character generally identified by the numeral 10. Contiguous character 10 includes an interior portion 12 and edges 14. Edges 14 are continuous and form the outer boundary of contiguous character 10.
FIG. 2 illustrates a dot matrix print character, generally identified by the numeral 16. Dot matrix print character 16 includes a plurality of dots 18 used to form the character. Each of the dots 18 include an interior portion 20 and an edge portion 22.
Referring to FIG. 3, a block diagram of the present dot matrix print detector is illustrated and is generally identified by the numeral 24. Dot matrix print detector 24 includes an optical character scanning device 26 which responds to light in the visible, infrared, X-ray or ultraviolet spectrum and preferably responds according to the human eye or photopic response. Scanner device 26 may include, for example, a CCD self-scanned array. The output of scanner device 26 provides a signal proportional to the reflectivity of the document at each sample or picture element (pixel) point.
The output of scanner device 26 representing raw video data in the form of pixel values is supplied to a high pass filter 28. High pass filter 28 allows energy of the dots 18 which compose a dot matrix print character 16 to pass while attenuating the energy of low frequencies associated with a contiguous character (FIG.1). High pass filter 28 may comprise, for example, a finite impulse response (FIR) or an infinite impulse response (IIR) filter. The output of high pass filter 28 representing the strength of the edges 14 or 22 of a character 10 or 16, respectively, is then totalled to represent the total energy level of edges of an area which may contain a character on the document.
The total energy level of the high pass filtered signal from high pass filter 28 for the given area is computed in accordance with Parseval's theorem in which the energy value for a particular pixel, row and column, is squared and the sum of the squared pixel energy values for the desired area are summed. Therefore, the output of high pass filter 28 is applied to a programmable read only memory (PROM) 30 which functions to square the pixel energy level output by high pass filter 28. The output of PROM 30 is applied to an accumulator 32 which functions to accumulate the individual pixel energy levels for the desired area of interest. The output of accumulator 32 is applied to a programmable read only memory (PROM) 34 which is a decision PROM for determining the presence of dot matrix print in the area of interest.
PROM 34 makes a comparison between the energy level of the high pass filtered signal with an energy level from an arbitrary filter such as, for example, a band stop filter, no filter at all or as shown in the FIGURE, a low pass filter 36. Low pass filter 36 may comprise, for example, a FIR or IIR filter functions to pass energy associated with the interior portions 12 or 20 of characters 10 or 16, respectively. These energy levels are applied to a PROM 38 which squares the energy level per pixel. The output of PROM 38 is applied to an accumulator 40 for summing the energy levels associated with the area of interest. The output of accumulator 40 is applied to PROM 34.
The size of the desired area selected may depend upon a number of factors such as, for example, the number of points per inch in the image of the document. If the area is selected too large, the detector will respond too slowly, whereas if too small an area is selected, the detector may operate incorrectly in the area between characters or in the presence of noise. An area of 32 by 45 or 32 by 60 pixels for a 300 point per inch image is representative of an area that can be used with the present detector 24.
A reset signal is applied to both accumulators 32 and 40 in order to reset the summing operation performed by accumulators 32 and 40 when a new area of interest is scanned by scanner device 26.
PROM 34 functions to output a signal indicating the presence of dot matrix print in the area of interest. PROM 34 may function to calculate a ratio between the energy sum signal output from accumulator 32 to the energy sum signal output by accumulator 40. This ratio can then be compared to a reference value, and if the ratio exceeds the reference value, an output signal is generated by PROM 34 such as, for example, a high or "one" signal indicating the presence of dot matrix print in the area of interest. If the ratio is less than the reference value, a low signal or "zero" output is generated by PROM 34 indicating that a contiguous character is present in the area of interest or that a decision cannot be made. Alternatively, PROM 34 may calculate an energy level representing the difference between the energy level output by accumulator 32 and the energy level output by accumulator 40 and compare this difference value to a reference value. If the ratio or difference signal exceeds the threshold this determination indicates that there is sufficient signal for a potential character to be present in the area of interest. The output of PROM 34 indicates the presence of dot matrix print in the area of interest.
High pass filter 28 may comprise, for example, a Texas Instruments model TMS320; low pass filter 36 may comprise, for example, a Texas Instruments model TMS320; PROMs 30 and 38 may comprise, for example, an AMD model AM90C644; accumulators 32 and 40 may comprise, for example, a Texas Instruments model LS181; and PROM 34 may comprise, for example, an AMD model AM99C641.
It therefore can be seen that the present dot matrix print detector performs a comparison between the high pass energy levels and the low pass energy levels of a character. The total energy levels of pixels in an area of interest is calculated, and based upon a decision function, a determination is made as to whether dot matrix print is located in the area of interest.
Whereas the present invention has been described with respect to a specific embodiment thereof, various changes and modifications may be suggested to one skilled in the art, and it is intended that the present invention encompass such changes and modifications as fall within the scope of the appended claims.
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US5121440 *||Aug 30, 1990||Jun 9, 1992||Monolithic Resources Corporation||Analog video character recognition system|
|US5357581 *||Nov 1, 1991||Oct 18, 1994||Eastman Kodak Company||Method and apparatus for the selective filtering of dot-matrix printed characters so as to improve optical character recognition|
|US5392365 *||Dec 23, 1991||Feb 21, 1995||Eastman Kodak Company||Apparatus for detecting text edges in digital image processing|
|US5394482 *||Jan 4, 1994||Feb 28, 1995||Eastman Kodak Company||Method and apparatus for the detection of dot-matrix printed text so as to improve optical character recognition|
|US5712922 *||Nov 15, 1994||Jan 27, 1998||Eastman Kodak Company||Neural network optical character recognition system and method for classifying characters in a moving web|
|US6100858 *||Sep 30, 1997||Aug 8, 2000||Infineon Technologies North America Corp.||Alphanumeric display with 21-dot matrix format|
|EP0534193A2 *||Sep 3, 1992||Mar 31, 1993||Eastman Kodak Company||Method for detecting ink jet or dot matrix printing|
|EP0534193A3 *||Sep 3, 1992||Apr 13, 1994||Eastman Kodak Co||Title not available|
|EP0539853A2 *||Oct 21, 1992||May 5, 1993||Eastman Kodak Company||Method and apparatus for locating and filtering dot-matrix video for improved optical character recognition|
|EP0539853A3 *||Oct 21, 1992||May 4, 1994||Eastman Kodak Co||Title not available|
|EP0539854A2 *||Oct 21, 1992||May 5, 1993||Eastman Kodak Company||Method and apparatus for the detection of dot-matrix printed text so as to improve optical character recognition|
|EP0539854A3 *||Oct 21, 1992||May 4, 1994||Eastman Kodak Co||Title not available|
|EP0552704A2 *||Jan 18, 1993||Jul 28, 1993||Eastman Kodak Company||Processing of dot-matrix/ink-jet printed text for optical character recognition|
|EP0552704A3 *||Jan 18, 1993||May 25, 1994||Eastman Kodak Co||Processing of dot-matrix/ink-jet printed text for optical character recognition|
|U.S. Classification||382/176, 382/182|
|International Classification||G06K9/20, G06K9/68|
|Cooperative Classification||G06K9/00442, G06K9/6807|
|European Classification||G06K9/00L, G06K9/68A|
|Jun 5, 1987||AS||Assignment|
Owner name: RECOGNITION EQUIPMENT INCORPORATED, 2701 E. GRAUWY
Effective date: 19870604
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST.;ASSIGNOR:WEIDEMAN, WILLIAM E.;REEL/FRAME:004722/0442