Method for data compression for two-value picture image

4,566,128

1 2

segments as an aggregate of the contour lines, and furMETHOD FOR DATA COMPRESSION FOR ther compressing the data of the original picture image

TWO-VALUE PICTURE IMAGE by removing any redundant coordinate point data

through comparison of the gradients of the closed-loop This invention relates to a method for compressing 5 line segments for each neighboring coordinate points, data of a two-value picture image and, in particular, to The improvement of this invention over the abovea method for compressing digital data for character font mentioned copending U.S. patent application exists in in computerized photo-type-setting. the elimination of any redundancy in the data and this

In computerized photo-type-setting, it is desirable to may be conveniently accomplished by eliminating those have various font sets such as Ming-type, gothic and so 10 coordinate points which have the same gradients as the on in the case of kanjis (Chinese characters) in memory preceding coordinate points along the contour lines, in readily retrievable manner. Also, to meet the need for Now this invention is described in the following in high quality, the bit pattern of each character must be terms of a specific embodiment with reference to the comprised of an extremely dense dot matrix, as opposed appended drawings, in which:

to the case of character patterns for common kanji pro- 15 FIG. 1 is a block diagram of a device for data comcessors. pression adapted to carry out the method of this inven

It thus gives rise to the problem of the need for an tion; extremely large memory device when one tries to store FIG. 2 is a bit pattern in the memory showing a typia number of character patterns on the order of at least cal original picture image whose graphic data is to be several thousand as character data which accurately 20 compressed;

represent the original character patterns. FIG. 3 shows in detail two mutually adjacent partial

There have been proposed various methods for com- picture images of the original picture shown in FIG. 2 pressing character data instead of simply storing each as the bit patterns stored in the corresponding memory character as a simple dot matrix and most of them were blocks;

based on the bandwidth compression techniques devel- 25 FIG. 4 is a block diagram showing a concrete strucoped in the field of image transmission technology. ture of the gradient comparison/elimination means of

However, the achievements so far made are by no FIG. 1; means impressive. For instance, the known run-length FIGS. 5 and 6 show a flow chart illustrating the method can only achieve a compression ratio of 20 to action of the data compression device shown in FIG. 1. 30%. And the vector method which can achieve a com- 30 FIG. 1 is a block diagram showing an embodiment of pression ratio on the order of a few percent required so a device for carrying out the method of this invention much manual work that it is almost inapplicable to for compressing the data of a two-value picture image, computer-controlled photo-type-setting. The original picture pattern whose graphical data is

As a break-through in this technical field, the Inven- to be compressed is photoelectrically scanned in a suittors disclosed a data compression method in U.S. Pat. 35 able manner and converted into a picture signal by an application No. 440,559 (which corresponds to Japa- original picture input device 1, and the picture signal is nese Patent Application No. 56-180649 filed on Nov. 11, converted into a digital signal by an A/D converter 2 1981), now U.S. Pat. No. 4,524,456, as a solution to the and then supplied to a two-value picture image memory above-mentioned problem, which can provide an im- device 3.

proved compression ratio over the conventional run- 40 The data of the original picture pattern stored in the length method or the vector method, facilitating the two-value picture image memory device 3 is comprised filing of two-value picture images and allowing sub- of a bit pattern of a dot matrix as illustrated in FIG. 2, stantial reduction in the necessary memory capacity by and the original picture pattern which expressed by the the data compression of a very high degree. bit pattern is the two-value picture image (which is

This invention is an improvement of the invention 45 referred to as "original picture image" hereinafter) disclosed in the above-mentioned patent application and which is to be compressed according to this invention, its primary object is to provide a method which is The two-value picture image memory device 3 is adapted to compress data at a very high ratio by detect- comprised of a plurality of memory blocks Moo~M,y ing the characteristics of the contour lines of partial arranged in a matrix, and each of the memory blocks picture images, which are divided up from the original 50 Moo~M,y stores one of the ixj partial picture images picture image, over the neighboring partial picture im- divided up from the original picture image, ages. Then, each partial picture image is transferred from

According to this invention, such a technical object is the corresponding memory block Moo ~ My to a partial accomplished by providing a method for data compres- picture image memory device, sequentially and for each sion for two-value picture image, characterized by the 55 memory block, according to the control command from steps of dividing the original picture image into a plural- a memory control device 5.

ity of partial picture images, determining a start point The memory control device 5 controls the writing, and an end point of each contour line found in each of reading and addressing of data in regards to both the the partial picture images by finding the intersections memory devices 3 and 4 according to the procedure between the contour line and the lines dividing the 60 given by a flow chart hereinafter, original picture image into the partial picture images, FIG. 3 is a magnified view of the bit patterns of two

detecting characteristic points in each contour line by partial picture images which have been transferred to tracing the contour line in each of the partial picture the partial picture image memory device 4, and the ones images from the corresponding start point and end shown in the drawing corresponds to the bit patterns of point, extracting those start points and end points which 65 the memory block M03 and M13 of FIG. 2. connect to those in the neighboring partial picture im- According to this bit pattern in the partial picture ages and the characteristic points as necessary coordi- image memory device 4, the pixel bits on the intersecnate points for determining a group of closed-loop line tions between the contour lines of the partial picture image and the lines dividing the pertinent partial picture plied to a data connecting means 10 for between the

image from the neighboring partial picture images as partial picture images so that the end point and the

start points P<s> and end points P(£). start point P'5), which are to be mutually connected

In other words, the outer periphery of the addresses between the contour lines which are separated one from

of the partial picture image memory device is sequen- 5 the other as separate partial picture images are con

tially designated from the origin (0,0) in counter-clock- nected together and is converted into closed loop line

wise direction and the changes of the pixel bits on the segment data consisting as a whole of several closed

outer periphery of the bit pattern are determined. loops.

In the bit pattern here, those bits in the character The signal processing described thus far is identical to

region are expressed by "1" while those in the back- 10 the one shown by the flow chart of FIG. 5 and has been

ground region are expressed by "0". In other words, the already discussed in detail in previously mentioned

pixel bits on the outer periphery of the addresses in the Japanese Patent Application No. 56-180649.

partial picture bit pattern are sequentially read out in The essential part of the method of data compression

counterclockwise direction and the bit corresponding according to this invention is found in the connecting

to "1" when the read-out bit is about to change from 15 means 10 for between different partial picture images

"1"1 to "0" is determined as a start point P® while the shown in FIG. 1 and thereafter.

bit corresponding to "1" when the read out bit is about At the connecting means 10 for between different

to change from "0" to "1" is determined as an end point partial picture images, the start point PW and the end

P'^. point Pt^) corresponding to each of the line segment

The contour line of the bit pattern generally extends 20 data are connected together so that the contour line for

on the periphery of the pixels, but, to simplify the fol- the whole original picture image may be reproduced

lowing description, those pixels or a row of pixels hav- and written into a closed loop line segment group mem

ing a contour line at least on one end are assumed to ory device 11 after being converted into a plurality of

make up a contour line. closed loop line segments.

The coordinate values (x„ y,) of the start point P<-^ 25 The original picture pattern and the contour line of its

and the end point P(^> detected by a start/end point bit pattern are each comprised of a closed loop as shown

detecting means are stored in an appropriate register by an example given in FIG. 2.

and supplied to a contour line tracing means 7 so that a Each closed-loop line segment data is then supplied contour line may be detected with the terminal points to a gradient comparison/elimination means 12 so that defined by the signal from the register. 30 the gradients of the pixel bits lying on the lines defining The pixel bits on the contour line traced by the con- the partial picture images and the contour lines with tour line tracing means 7 are given to a characteristic respect to the neighboring coordinate points are depoint detecting means 8 so that the changes in the char- tected and those pixel points having the same gradients acteristics of the contour line may be detected and ap- ahead and behind are eliminated from the coordinate propriate number of characteristic points on the con- 35 point data.

tour line may be sequentially detected. Specifically, the gradient comparison/elimination

These characteristic points may be classified, for means 12 may be comprised of, for instance, three data

instance, in the following manner according to the type buffer memories 21, 22 and 23, two gradient computing

of the changes in the contour line. circuits 24 and 25, a comparison circuit 26 and a gate

PC): point at which the direction of the contour line 40 circuit 27.

changes abruptly The coordinate point data of a certain closed loop

P<2>: point at which the direction of the contour line line segment group is sequentially stored in the data

changes gradually in one direction buffer memories 21, 22 and 23 as coordinate points P«,

P(3>: point at which the contour line changes from a P„ + l, P„ + 2, • • ■ from the memory device 11 for the

straight line to a curved line 45 closed loop line segment group under the control com

P<4>: point at which the contour line changes from a mand from a closed-loop tracing control circuit 13.

curved line to a straight line The data buffer memories 21 and 22 are connected to

The group consisting of the coordinate points ob- the gradient computing circuit 24 and the data buffer

tained by sequentially tracing the coordinate points of memories 22 and 23 are connected to the gradient com

these characteristic points from the start point to 50 puting circuit 25. And the gradient between the coordi

the end point P^ are written into a line segment data nate points P„ and Pn+i is computed at the gradient

memory device 9 as the compression data of the partial computing circuit 24 while the gradient between the

picture image. coordinate points P„+i and Pn+2 is computed at the

Each of the coordinate points in the line segment data gradient computing circuit 25 so that the gradients thus

memory device is converted into the coordinate point 55 obtained are supplied to the comparison circuit 26.

of the original bit pattern according to the block num- When the gradients of P„P„+i and ... not

ber 00—ij of the memory block MrjO'-M//. in agreement, the output of the comparios circuit 26 is

In other words, for each of the memory blocks supplied to the gate circuit 27 and the output data of the Moo~M,y, the characteristic points p(')-~p(4) and the data buffer memory 22 storing the coordinate point associated start and end points PW and P(£) located at 60 P„+i is sent to the compressed data memory device 14. respective ends of the characteristic points are written Conversely, when the gradients of P„Pn+i and into the line segment data memory device 9, normally, P„+iP„+2 are in agreement, the data corresponding to in the order of the start point PW, the characteristic the coordinate point P„+i is not supplied to the cornpoints P(')~p(4)j ancj the end point P(£) as the com- pressed data memory device 14.

pressed data. 65 For instance, the processing in the gradient cornAfter all the compression data of the line segment parison/elimination means 12 in regards to the data data memory device 9 is collected in regards to all the P„(S), Pn+i(G), P,,+2(^, P«+3(2), P„+4(£). ... of the memory blocks Moo ~ My, the compression data is sup- closed loop line segment group over the memory blocks