JPH04588A - Method for recognizing character - Google Patents

Abstract

PURPOSE:To make it unnecessary to use a dictionary and to shorten a collating time by detecting the inclination angle of a character pattern, extracting peripheral distribution, detecting a division point, setting up a division boundary in a character circumscribed frame, extracting a feature in each divided area, and then stabilizing the extracted feature. CONSTITUTION:In a plane expressed by an X-Y coordinate system including a character pattern stored in a pattern register 4, coordinates applying the maximum and minimum values of alphax + betay relating to black picture elements (BPEs) in the pattern are detected by a character inclination detector 8. A peripheral distribution extracting part 9 extracts peripheral distribution based on an inclination angle and a division point detecting part 10 detects a division point by using the distribution. Then a feature matrix extracting part 11 sets up a division boundary based on the inclination angle and the peripheral distribution and extracts a feature matrix. Thereby, the feature extracted even from an inclined character pattern can also be stabilized, so that the preparation of a dictionary corresponding to the deformation of character inclination in an identification (ID) part is unnecessary. Thus, a dictionary collating time can be shortened and a processing time can be increased.

Description

Detailed Description of the Invention (Field of Industrial Application) The present invention relates to a character recognition method that recognizes input characters from a character pattern obtained by photoelectrically converting characters on a medium, which is particularly fast and has good recognition accuracy. It concerns a character recognition method.

(Prior art) 5 conventional character recognition methods of this type are Japanese Patent Application Laid-open No. 58-1.
There was one described in Publication No. 23171.

Conventionally, when recognizing characters on a medium, the extracted features vary due to the moving speed and slope of line elements due to differences in the writer in handwritten characters, and differences in font (font) in printed characters. , it was necessary to prepare a dictionary that accommodates changes in the characteristics. As a result, the capacity of the four dictionaries has increased, and the time required to collate the dictionaries has also increased, resulting in a decrease in processing speed and an increase in the size of the device.

Therefore, in the technique of the above-mentioned document, the character pattern is scanned in the horizontal and vertical directions to obtain the peripheral distribution, and the dividing coordinates are determined based on the position of the center of gravity of the peripheral distribution. This makes it possible to make the dividing position for determining the area on the character stamp follow the movement of the line element position, absorbing fluctuations in characteristics such as stroke position fluctuations based on individual differences, and absorbing fluctuations in characteristics such as stroke position fluctuations due to individual differences, and We are trying to speed up the processing speed.

(Problems to be Solved by the Invention) However, the above character recognition method has the following problems.

Figures 2 (a) and (b) are diagrams for explaining the conventional character recognition method. The character patterns and their surrounding distribution are shown respectively.

In the conventional character MX recognition method, an input character pattern is scanned in the horizontal and vertical directions to obtain a peripheral distribution, and division coordinates are determined based on the position of the center of gravity of the peripheral distribution. However,
As shown in Figure 2 (b), when the marginal distribution is extracted for an italic character whose vertical line elements have a slope, Figure 2 (a
Compared to the case of a character with no slope as shown in ＞, the density distribution in the vertical line element portion becomes gentler, so the detected division point coordinate values are different. Moreover, characters with no slope are divided parallel to the vertical line elements,
For italic characters with vertical line elements that are slanted, the character lines are divided diagonally. Therefore, even though the characters have the same shape, the extracted feature amounts are significantly different, resulting in a decrease in character recognition accuracy.

Therefore, in conventional character recognition methods, dictionaries that accommodate various transformations have been prepared in order to absorb such variations in characteristics. As a result, there were problems in that the dictionary capacity increased and collation took time, which increased the hardware scale (equipment scale) and processing time, and it was not possible to obtain a solution that was technically satisfactory.

The present invention provides a character recognition method that solves the problems of the prior art, such as an increase in processing time and an increase in the size of the device.

(Means for Solving the Problems) In order to solve the above problems, the present invention detects a circumscribing frame of a character pattern obtained by photoelectrically converting characters on a medium, and divides the inside of the circumscribing frame into a plurality of regions. After extracting features for each divided region, the character recognition method recognizes the characters by comparing the extracted features with standard features prepared in advance, and the following measures are taken. It is something.

That is, in the plane defined by the X-Y coordinate system including the character pattern, αχ + βy regarding the black pixel of the character pattern for at least two sets of specific real numbers α and β (however, y is The coordinates giving the maximum and minimum values of the coordinate values) are determined, and the inclination angle of the character pattern is detected based on the coordinates. Next, the character pattern is scanned based on the inclination angle to extract a peripheral distribution, and dividing points within the circumscribed frame are detected using the peripheral distribution. Thereafter, a dividing boundary within the circumscribing frame is set based on the inclination angle and dividing point, and the inside of the circumscribing frame is divided into a plurality of regions based on the dividing boundary.

(Function) According to the present invention, since the character recognition method is configured as described above, when a character pattern is input, the circumscribing frame for the character pattern is detected, and the maximum of αX+βy regarding the black pixels of the character pattern is detected. The coordinates giving the value and the minimum value are detected, and the tilt angle of the character pattern is detected based on the coordinates. Then, a peripheral distribution is extracted according to the detected inclination angle, and dividing points within the circumscribing frame are detected using the peripheral distribution.

Next, a dividing boundary within the circumscribing frame is set based on the detected inclination angle and dividing point, and the inside of the circumscribing frame is divided into a plurality of regions based on the set dividing boundary.

Thereafter, the input characters can be recognized by extracting features for each divided region and comparing the features with standard features.

In this way, the inclination angle of the character pattern is detected, and based on that inclination angle, the character pattern is scanned and the marginal distribution is extracted, so character patterns that are inclined with respect to the vertical axis can also be extracted. The characteristics of the characters are stable, and there is no need to prepare a dictionary that accommodates the deformation of the character slant. This makes it possible to increase the processing speed, reduce the size of the device, and improve recognition accuracy. Therefore, the above problem can be solved.

(Embodiment) FIG. 1 shows an embodiment of the present invention, and is a functional block diagram of a character recognition device for explaining a character recognition method.

This character recognition device has a photoelectric conversion unit 1 that converts an optical signal IN of a character image on a medium such as a form into a quantized electric signal (digital signal), and a line buffer 2 is provided on the output side of the photoelectric conversion unit 1. It is connected. Row buffer 2 has a width of 20, for example.
It has a size of 48 bits x height of 128 bits, and is configured to store the digital signal of one line of character image, and the output side is connected to a button register 4 via a character cutting section 3.
is connected.

The character cutting unit 3 has a memory, and receives one character's worth of digital signals from the output of the line buffer 2 (this is called a "character pattern").
) is stored in the button register 4. The button register 4 has a storage capacity of, for example, 64 x 64 bits, and is connected to its output side with a character frame detection section 5, a line width measurement section 6, and a sub-button extraction section 7, as well as a character slant detection section 8 and A peripheral distribution extraction section 9 is connected.

The character frame detection unit 5 has a function of scanning the character pattern in the button register 4 to detect a circumscribed frame, that is, a character frame, and providing the detection result to the character slope detection unit 8, the peripheral distribution extraction unit 9, etc. ing. The line width measuring section 6 has a function of measuring the line width of the output of the button register 4 and providing the measurement result to the sub-button extracting section 7. The sub-battle extractor 7 scans the button register 4 in a plurality of directions to extract vertical, horizontal, right diagonal, and left diagonal sub-battens, and includes a vertical sub-pattern extractor 7a, a horizontal sub-pattern extractor 7b, and a right diagonal sub-battle extractor 7c. , and a left diagonal sub-button extraction section 7d. Each of the extraction units 7a to 7d has a memory for storing a button.

The character slant detection section 8 has a function of extracting the slant of the character pattern stored in the button register 4.

The peripheral distribution extraction section 9 has a function of scanning the slam register 4 using the slope output from the character slope detection section 8 and extracting the peripheral distribution. A dividing point detecting section 10 is connected to the output side of the peripheral distribution extracting section 9, and the output sides of the dividing point detecting section 10, character frame detecting section 5, sub-butter extracting section 7, and character slope detecting section 8 are as follows. Feature matrix extraction unit 1
1 is connected, and an identification unit 12 is further connected to its output side.

The division point detection unit 10 has a function of detecting division point coordinates for dividing the inside of the circumscribing frame into a plurality of partial regions using the peripheral distribution extracted by the peripheral distribution extraction unit 9. The feature matrix extraction unit 11 extracts feature amounts from each of the vertical, horizontal, right diagonal, and left diagonal sub-patterns output from the sub-battle extraction unit 7 to create a feature matrix,
It has a function of providing it to the identification section 12. Identification part 1
2 is the standard character feature matrix (standard character mask) G
(k) and the character names of standard characters having this feature matrix G(k). Then, the feature matrix F(k) extracted by the feature matrix extraction unit 11 and the feature matrix G (
k), the feature matrix F
It has a function of recognizing the character/figure in the circumscribed frame area obtained in (k) and outputting the character name ○LJT.

FIG. 3 is a functional block diagram showing an example of the configuration of the character slant detection section 8 of FIG. 1. As shown in FIG.

This character inclination detection section 8 includes an X coordinate generating means 21 for assigning an X coordinate to a pixel of image data M input from the button register 4, and a Y coordinate generating means 21 for assigning a Y coordinate to a pixel of image data M. generating means 22, and coordinate detecting means 30, 40 are connected to their output sides. The coordinate detection means 3040 detects the maximum and minimum calculated values αX+βy regarding the pixels of the extracted pattern using the X, Y coordinates and at least two sets of specific α values and data values.

X2 of the pixel of the extracted pattern that gives the maximum and minimum calculated values
It outputs each Y coordinate as a feature point coordinate, and the first rule is the inclination that calculates the inclination of a character pattern based on the feature point coordinates! Output means WOO is connected.

The coordinate detecting means 30 includes an X0Y calculating means 31 for calculating the calculated value αχ+βy, a maximum value detecting means 32 for detecting the maximum calculated value, and a means for storing the X and Y coordinates of the pixel giving the maximum calculated value. A maximum value coordinate storage means 33, a minimum value detection means 34 for detecting the minimum calculated value, and a minimum value coordinate storage means 35 for storing the X and Y coordinates of the pixel giving the minimum calculated value. has been done. Similarly, the coordinate detection means 40 includes an x-y calculation means 41, a maximum value detection means 4
2, maximum value coordinate storage means 43, minimum value detection means 44, and minimum value coordinate storage means 45.

In this character inclination detection section 8, in order to detect feature point coordinates,
For example, two sets of α: α=β=1 and α==−1, β=−1
Since the value and β value are used, two coordinate detection means 30.4
0. In one coordinate detection means 30, the calculated value
The other coordinate detecting means 40 has a function of detecting the coordinates of the pixel that provides the maximum and minimum calculated values for +Y, and the other coordinate detection means 40 provides the maximum and minimum calculated values for the total X value X-Y.

For example, the X+Y calculation means 31 is an addition circuit, the X-Y calculation means 41 is a subtraction circuit, and the detection means 32° 34.42.4
4 are a comparator and a register, respectively, and coordinate storage means 33
, 35, 43, and 45 are each composed of registers and the like. In addition, the X coordinate generating means 21 and the Y coordinate generating means 2
2 is configured to be used in common with the coordinate detection means 30.40, and therefore the coordinate detection means 30.40
has a function of detecting feature point coordinates by inputting the X, )' coordinates output from the X coordinate generating means 21 and the Y coordinate generating means 22, respectively.

Next, we will explain the processing of each functional block (
I) to (■) will be explained.

FIG. 4 is a diagram showing an example of printed characters in italics. In this embodiment, as shown in FIG. 4, a character including a character in which the right and left sides of a quadrilateral obtained by connecting four feature points of a character are inclined, such as an italic character in a printed character. The case of recognizing columns will be explained below.

(I> When the optical signal IN of the character image written on the character pattern generation processing form is input to the photoelectric conversion unit 1, the optical signal IN is input to the photoelectric conversion unit 1.
The digital signal of the value, that is, the character line part is converted to "1" and the background part is converted to 'O'.The digital signal of the character image for one line converted by the photoelectric conversion unit 1 is stored in the line buffer 2. .

The character cutting section 3 reads out a digital signal (character pattern) for one character from the digital signal of the character image stored in the line buffer 2, and stores it in the pattern register 4. In this embodiment, the form format is specified in advance, and the address indicating the character position in the line buffer 2 is stored in the memory L of the two-character extraction unit 3.Therefore, the character extraction operation is specified using the address. This is executed by reading out the contents of the row buffer 2.

(ff) Character frame detection/line width measurement processing The character frame detection unit scans the pattern of the button register 4 and obtains the left end coordinates X, il, right end, mark Xr, top end coordinate Yt, and bottom end coordinate yb of the pattern. To detect. The circumscribing frame, or character frame, is (
, XJ), Yt), (XJII, Yb), (Xr,
A rectangular frame connects the four points Yt), (Xr, Yb).
Calculate the size of the required character frame. That is, the size of the character frame in the parallel direction (horizontal direction) to the X-axis of the button register 4 is WPh = Xr - XR ÷ 1, and the size of the character frame in the perpendicular direction (vertical direction) is WPh. W as WPv
P v = Y t - Yb+1 is calculated for h, respectively. Further, WRh+WPv Z is calculated by setting the size of the character frame in the diagonal right direction and the diagonal left direction 45' as WPr and WPN.

On the other hand, the line width measurement unit 6 inputs the digital signal from the button register 4 and counts, for example, the number Q of states in which all points in a 2×2 window are black bits and the number A of all black bits. Then, the line width WL is calculated according to the conventionally known equation (1).

WL=A/ (A-Q) ・・・・・・(
1) (III) Sub-battle extraction process In the sub-battle extraction unit 7, the vertical sub-battle extraction unit 7a,
Horizontal sub-pattern extraction section 7b, right diagonal sub-pattern extraction section 7
C and the left diagonal sub-butter extraction unit 7d, respectively, are set on the button register 4 in a direction perpendicular to the X-axis direction (vertical direction), a direction parallel to the X-axis direction (horizontal direction), and a direction 45 degrees counterclockwise from the X-axis. (45° diagonal right direction) and 45° clockwise diagonal direction (45° diagonal left direction) are the main scanning directions, and the button register 4 is scanned, and the vertical, horizontal, right diagonal, and left Extract diagonal sub-battens.

That is, the vertical sub-pattern extraction unit 7a scans the entire ring pattern with the vertical direction as the main scanning direction, and detects continuous black bits (black runs) on the vertical scanning line. and,
The length p that satisfies the following formula (2) from among the detected black runs
Extract the black orchid.

1≧N, WL (2) However, p: length of the black run in the main scanning direction N: an arbitrary constant for each sub-pattern (for example, 2) Next, the vertical sub-pattern extraction unit 7a extracts ( 2) A black run that satisfies the formula is regarded as a black run constituting one sub-pattern and is stored in a vertical sub-pattern memory (not shown). Black runs that do not satisfy equation (2) are regarded as white bits.

Similarly, the horizontal, right diagonal, and left diagonal sub-button extraction parts 7b
, 7c, and 7d scan the ring pattern with the horizontal, diagonal right, and diagonal left directions as the main scanning directions, and select black runs that satisfy equation (2>) from among the black runs on the scanning lines in the respective main scanning directions. The extracted black runs are stored in horizontal, right diagonal, and left diagonal sub-pattern memories (not shown) as black runs constituting a sub-pattern.

(IV) Character slant detection processing The character slant detection section 8 extracts the slant of the character slams stored in the button register 4. This extraction method
Figure 3, Figure 5 (a>, (b) and Figure 6 (a>, (b)
).

FIGS. 5(a) and 5(b) are diagrams showing an example of detecting feature points to explain the slope detection method.On the plane expressed in the X-Y coordinate system, there is a button M2 to be extracted. Characters with no slope (FIG. 5(a)) and italic characters (FIG. 5(b)) are shown. BL, BR, TL, and TR are feature points. FIG. 6(a).

(b) explains the character slant detection method;
FIG. 11A is a flowchart of maximum value processing, and FIG. 10B is a flowchart of minimum value processing.

This character inclination detection process will be explained below by dividing it into (A>detection process of feature point TL and BR coordinates, (B) detection process of feature point BL and TR coordinates, and (C) inclination degree calculation process.

(A> Detection process of feature points TL and BR coordinates In this process,
The following (1) is detected by the coordinate detection means 30 shown in FIG.
This is executed in the processing steps of ~(5).

(1) Steps 81 to S3 First, scanning of the slam register 4 is started, and the image data M output from the slam register 4 is input to the character slant detection section 8 (step Sl).

This image data M is input to the maximum value detection means 32 and the minimum value detection means 34 in the coordinate detection means 30 for each pixel.

Along with this, the X coordinate generating means 21 and the Y coordinate generating means 2
2 generates χ and Y coordinates associated with each pixel of the image data M in synchronization with the output of the image data M. As a result, these generating means 21 and 22 assign X and Y coordinates to the image data M. and the output X
, Y coordinate is determined by the X+Y calculation means 31 in the coordinate detection means 30.
, maximum value coordinate storage means 33, and minimum value coordinate storage means 3
(Step S2).

When the X+Y calculation means 31 receives the X and Y coordinates, it calculates a calculated value X+Y from these χ and Y coordinates, and outputs the calculated value to the maximum value detection means 32 and the minimum value detection means 34 (step S3). Through steps 82 to S3, the maximum value detection means 32 receives the image data M and the calculated value, the minimum value detection means 34 receives the image data M and the calculated values, the maximum value coordinate storage means 33 stores the X, Y coordinates, and the minimum value coordinate. Storage means 35
Inputs the X and Y coordinates for each pixel. Then, the coordinate detection means 30 performs step S4 described in t★, Sea
, S5b or S7 is repeatedly executed for each pixel L-'', and processing is performed according to the result of the judgment.

(2> Step S4 The maximum value detection means 32 and the minimum value detection means 34 detect that the image of the input image data M has a predetermined pixel value (for example, '1
°).

(2) When the pixel does not have a predetermined pixel value, the maximum value detection means 32 and the minimum value detection means 34 do not rewrite the stored comparison values, and at the same time, the maximum value coordinate storage means 33 and the minimum value coordinate storage means 33 also Do not rewrite the stored X and Y coordinates. The detection means 32, 34 performs the determination in step S7 after step S4.

■ When the pixel value is a predetermined value, the maximum value detection means 32 compares the comparison value and the calculated value (step S''; a) after step S4, and the minimum value detection means 3
4, the comparison value and the calculated value are compared (step 55b) after step S4.

(3) Step S5a In this step, it is determined whether the calculated value is greater than the comparison value, and the following processing is executed depending on the determination result.

■ When the calculated value>comparison value, the maximum value detection means 32 stores a calculated value larger than the comparison value as a new comparison value instead of the previously stored comparison value (rewriting the comparison value), and at the same time The set pulse is output to the maximum value coordinate storage means 33. The maximum value coordinate storage means 33 into which the set pulse is input stores the stored X,
Instead of the Y coordinate, the X and Y coordinates of the pixel that gives a calculated value larger than the comparison value are newly stored ("Rewriting X and Y coordinates" step S6).

■ When the calculated value≦comparison value, the maximum value detection means 32 stores the previously stored comparison value without rewriting it, and the maximum value coordinate storage means 33 also rewrites the stored X and Y coordinates. do not have.

The maximum value detecting means 32 performs the determination in step S7 after step S5a in both cases (1) and (2).

Note that as the initial value of the comparison value stored in the maximum value detection means 32, a value smaller than the value that can be taken as the calculated value α may be used, for example. For example, the image data M is divided into p rows and columns of pixels with α=β−1 (therefore, ○≦X≦m
−1 and 0≦Y≦j−1), −1 can be set as the initial value of the comparison value. Alternatively, the calculated value αχ1βy inputted first to the maximum value detection means 32 may be used as the initial value of the comparison value.

Further, the maximum value detection means 32 rewrites the comparison value and the X, Y coordinates when the calculated value αχ + βy is larger than the comparison value and when the calculated value is equal to the comparison value, and at the same time When the value is smaller than the comparison value, the comparison value and the X, Y coordinates may not be rewritten. Furthermore, any suitable numerical values may be used as the X and Y coordinates of the maximum value coordinate storage means 33.

(4) Step S5b In step S5b of FIG. 6(b), it is determined whether the calculated value is smaller than the comparison value, and the following processing is executed based on the determination result.

■ When the calculated value is a comparison value, the minimum value detection means 34 stores a new calculated value smaller than the comparison value as a comparison value instead of the previously stored comparison value (rewrites the comparison value), and The minimum value coordinate storage means 35 outputs a set pulse. When the set pulse is input, the minimum value coordinate storage means 35 stores the stored
, Y coordinates are rewritten to the X, Y coordinates of the pixel that gives a calculated value smaller than the comparison value (rX, Y coordinate rewriting J step S6).

■ When the calculated value≧comparison value, the minimum value detection means 34 stores the previously stored comparison value without rewriting it, and the minimum value coordinate storage means 35 also rewrites the stored X and Y coordinates. do not have.

The minimum value detection means 34 performs the determination in step S7 after step Sob in both cases (1) and (2) above.

Note that as the initial value of the comparison value stored in the minimum value detection means 34, a value larger than the value that can be taken as the calculated value αχ+βy may be used, for example. For example, α=β=1,
Image data M is divided into 1 row and m columns of pixels (therefore, 0
≦X≦m-1 and 0≦Y≦J-1), then
m+n-1 can be set as the initial value of the comparison value. Alternatively, the calculated value αχ+βy inputted first to the minimum value detection means 34 may be used as the initial value of the comparison value.

Further, the minimum value detection means 34 rewrites the comparison value and the χ and Y coordinates when the calculated value is smaller than the comparison value and when the calculated value is equal to the comparison value, and at the same time, the calculated value is smaller than the comparison value. When the value is large, the comparison value and the χ and Y coordinates may not be rewritten. Furthermore, any suitable numerical values may be used as the X and Y coordinates of the minimum value coordinate storage means 35.

(5) Step S7 In step S7, it is determined whether or not scanning of the image data M has been completed, and the following processing is executed depending on the determination result.

■ When the scanning of the image data M is finished, the maximum value detection means 32 and the minimum value detection means 34 detect the
, Y coordinates are outputted to the maximum value coordinate storage means 33 and the minimum value coordinate storage means 35. Then, the coordinate storage means 33 and 35 output the stored X and Y coordinates as feature point coordinates. Together with this, the detection means 32 and 34 initialize the comparison value (step S8).

When processing is completed for all pixels, the X stored in the maximum value coordinate storage means 33 and the minimum value coordinate storage means 35
, Y coordinates of the pixel whose maximum and minimum calculated values are x, y
coordinates, that is, feature point coordinates.

In the coordinate detection means 30, α=β=1, so at the end of all processing, for example, the coordinates of the feature point BR of the extracted pattern M2 shown in FIG. 5 are stored in the maximum value coordinate storage means 33. , and the coordinates of the feature point TL are stored in the minimum value coordinate storage means 35.

■ When the scanning of the image data M is not completed If the scanning of the image data M is not completed and therefore the processing has not been completed for all pixels of the data M, the coordinate detection means 30 detects the remaining pixels of the image data M. Step 84. S5a, S5
b or S7 is made, and processing is performed according to the result of the judgment.

(B) Detection processing of feature point BL and TR coordinates This processing is performed by the coordinate detection means 40 and the like. That is, the coordinate detection means 40 performs almost the same processing as the coordinate detection means 30 in parallel with the operation of the coordinate detection means 30 described above.

In this coordinate detection means 40, α=1 and β=1, so when processing is completed for all pixels, for example, the fifth
The coordinates of the feature point TR of the extracted pattern M2 shown in the figure are stored in the maximum value coordinate storage means 43, and the feature point BL
The coordinates of are stored in the minimum value coordinate storage means 45.

(C) Slope calculation process In the processes (A) and (B), the feature points TR and TL.

Coordinates of the four points BR and BL (TRx, TRy).

(TLx, TLy), (BRx, BRy), (BLx,
When BLy) is detected, the inclination calculation means 50 calculates a value (inclination) TH corresponding to the inclination angle of the input character pattern from the coordinates of the four points according to the following equation.

That is, the inclination calculating means 50 uses equation (3) to calculate the difference between the X coordinate values of the upper and lower ends of the right side of the quadrilateral obtained by connecting the four feature points, and the X coordinate values of the upper and lower ends of the left side. The value obtained by dividing the average difference in coordinate values by the average height of one character is the slope TH (however,
TH; real number) and provided to the marginal distribution extraction section 9 and the feature matrix extraction section 11.

(V) Marginal distribution extraction process Figure 7 (a>, (b) is an explanatory diagram of the marginal distribution extraction method.
, Figure (b) shows the marginal distribution extracted from the pattern.

The peripheral distribution extraction unit 9 determines the scanning path of the character pattern shown in FIG. 7(a) according to the slope TH (=real number) obtained by the character slope detection unit 8. Figure 7 (a>, TH=
It is 5/20. Vertical scanning starts scanning from the top edge,
Coordinates (xl.) of the scanning path when starting from the start address (Xa, YT).

y·) can be expressed by the following equation (4).

xl=Xa y1=YT x ==Xa+T)lx (y, -YT)'5'i='
li 1+1 (4) However, the result of THX (yi-YT) is rounded down to the decimal point.

All coordinate values are integers.

The number of black bits in the scanning path is counted and taken as the value of the marginal distribution at Xa. It is varied within the range of XaliXL≦Xa≦XR. That is, the scanning start point is set to (XL.

YT>, (XL+1.YT),..., (XR
, YT), the number of black bits on the scanning path according to equation (4) is counted for each scanning start point, and the peripheral distribution is extracted. The extracted marginal distribution is 5X (x) (where x=X
L, ..., XR).

Applicable, scan in the horizontal direction and obtain peripheral distribution 5Y in the Y117 direction.
(y) (where y=YT, . . . , YB) is extracted.

In this embodiment, horizontal scanning is performed along a scanning path parallel to the X-axis. That is, when the pixel value at the coordinate (x, y) pixel of the pattern register 4 is P (x, y), it is calculated as follows.

(Vl) Division point detection processing In the division point detection unit 10, the marginal distribution SX (x>).

Using SX(,y), coordinates of division points on the X-axis and Y-axis for dividing the area within the circumscribing frame into NXxNY partial areas are determined for each circumscribing frame. However, NX is the number of divisions in the X-axis direction, and NY is the number of divisions in the YN direction.

The numbers of divisions NX and NY are preferably set to arbitrary suitable values depending on the complexity of the characters. For example, when recognizing characters with a small number of strokes such as kanji and katakana, (2X
2) Divide the area within the circumscribing frame into a small number of partial areas of about 3x3.

When kanji are to be recognized, (4X4) to (8X8
) The area within the circumscribing frame is often divided into partial areas of approximately However, in this embodiment, the area within the circumscribed frame is always divided into (4×4> pieces, for example) regardless of the complexity of the recognition target.

This division point detection unit 10 calculates
The division point coordinates DX (m>, DY (n)) are determined. It is found by dividing the sum of the next moments by the sum of black bits in that range.However, m and n are the barycenter coordinate numbers assigned in order of the size of the coordinate values, and m = 1 to
NX-1 (NX-1 is the number of centers of gravity in the X-axis direction), n=1
~NY-1 (NY-1 is the number of centers of gravity in the Y-axis direction).

First, target the range χL to XR of the character frame in the X-axis direction,
As shown in the following equation (5), by dividing the sum of the first moments of the peripheral distribution SX, (x>) of the input character pattern by the sum of black bits in that range, the barycenter coordinate DX ( Find 2>.

However, since the expression on the right side is a real number, Round up the decimal point DX Make (2) an integer.

Next, find the two centroids DX(1) and DX(3) for each range divided by the centroid coordinates DX(2>, XL to DX (2>-1, DX (2) to XR) .

DX(1) ......(6) DX(3) As above, the center of gravity coordinates DX(1), DX(2>, D
Find X (3) and determine the coordinates as the dividing point coordinates on the X axis.

Similarly, the coordinates of the center of gravity in the Y-axis direction are the peripheral distribution on the Y-axis 5Y (
y), the center of gravity coordinates DY(4) is detected for the range YT to YB of the character frame. Next, YT~Y (2>-1
, Y (2) ~YB, respectively, the marginal distribution 5
Detect the barycenter coordinates Y(1), Y(3) of Y(y>).

As described above, the center of gravity coordinates DY(1), DY(2>,
DY (3) is determined and the coordinates are determined as the division point coordinates on the Y axis.

The coordinates of the division points on the X-axis and Y-axis are given to the feature matrix 11.

(VN) Coordinates X where the character frame detection unit 5 defines the circumscribed frame of the character pattern
When J, Xr, Yt, and Yb are detected, and the dividing point detection section 10 further detects the target dividing point coordinates for the character pattern, the feature matrix extraction section 11 extracts each of the vertical, horizontal, right diagonal, and left diagonal sub-patterns. Extract features from and create a feature matrix.

That is, the feature matrix extraction unit 11 converts one circumscribed frame area into target division point coordinates and coordinates Xj, Xr.

It is divided into NXXNY partial regions by Yt and Yb, and the feature quantity representing the character line 1 of the subpattern in each partial region is extracted. Then, a feature matrix consisting of NXXNYX four feature quantities extracted from each sub-pattern within one circumscribed frame area is extracted as the feature quantity matrix of the circumscribed frame area.

First, feature extraction from the horizontal sub-pattern (H3P) will be explained.

The feature matrix extraction unit 11 extracts target division point coordinates and coordinates
The area within the circumscribing frame is divided into NXxNY partial areas based on fl, Xr, Yt, and Yb. The coordinates of the target dividing point and the coordinates XJ, Xr, Yt, and Yb are the dividing point coordinates),
For each partial area, the number of black bits BH(i) of the horizontal sub-pattern H8P within the partial area.

Count j〉.

The division area is determined from the division point coordinates as follows. X
The dividing point coordinates DX (n> on the axis) indicate the dividing point on the upper side of the circumscribing frame of the character pattern. Starting from this coordinate, the left side of the coordinate series obtained by the following equation (8) is the dividing boundary S. .

xl=DX(n>yl=YT x ,=DX (n>-THx (y = -YT)y
i=yi-1+1 (8) An example of this dividing boundary is shown in FIG. 7 (a>).

The horizontal division boundary S is set in the horizontal direction from DY (m'').

The number of black bits BH(i,,j) of the horizontal sub-pattern H8P is counted for each divided area divided as described above. This BH(i, j> is the number of black bits of the partial area in the i-th row and j-th column regarding one circumscribed frame area. Next, according to equation (9), Feature amount FH (
i, j).

However, i=1.2. ......, NXj=1.2.・
..., NY WL; line width wph, character width (=Xr-Xjl +1), and H
Similarly to the case of 8P, V of the partial area of the i-th row and j-th column is
The number of black bits BV (i, j>, B
Count R(i, j>, BL(i, j>) and use the following formula (10)
~ According to (12), v regarding the partial area in the i-th row and j-th column
sp.

H8P, LSP feature quantity FV(i, j>, FR(i, j
>, calculate FL(i,j).

However, WPv; character height (=Yb-Yt-i1)WPr=
WPN = (WP v + WP h > / 2 or more, VSP,
Extract the features of H3P, H8P, and LSP, and apply these NX
Feature matrix F (k
) (k=1.2...NXxNYx4) is obtained. The feature matrix extraction unit 11 extracts a feature matrix F (
k) is extracted for each circumscribed frame area, and the extraction result is sent to the identification unit 12.

(■) The identification processing identification unit 12 uses the extracted feature matrix F(k),
Feature matrix G (
k), the feature matrix F
Characters and figures in the area within the circumscribed frame for which (k> is obtained) are recognized. In this recognition, the feature matrix F is calculated according to the following equation (13).
Find the distance ti[) between (k> and G(, k), and find the character name of the standard character of the feature matrix G (k) where the distance is the minimum (for example, the character code specified in the JIS standard) OUT
is output as the recognition result.

As described above, this embodiment has the following advantages.

(a) In this embodiment, the character tilt detection unit calculates the coordinates giving the maximum and minimum values of αχ+βy regarding the black pixels of the character pattern in the plane expressed by the X-Y coordinate system that includes the character slam in the button register 4. 8, and further detects the inclination angle of the character pattern based on the coordinates. Then, a marginal distribution extracting section 9 extracts a marginal distribution according to the inclination angle, and a dividing point detecting section 10 detects dividing points using the marginal distribution. Thereafter, the feature matrix extraction unit 11 sets division boundaries based on the inclination angle and peripheral distribution, and extracts a feature matrix. Therefore, the extracted features are stable even for character patterns having an inclination. Therefore, there is no need to prepare a dictionary corresponding to the deformation of character slant in the identification unit 12, and the reduction in dictionary capacity can shorten the collation time, increase processing speed, and reduce the hardware scale, improving recognition accuracy. It can provide a good character recognition method.

(b) In the above embodiment, a method for recognizing italic characters in printed characters as shown in FIG. 4 has been described, but characters to be recognized are not limited to this. For example, in the law of handwritten characters, if the target is a character written upward to the right, the upper and lower sides of the quadrilateral obtained by connecting the four feature points of the character are slanted, and the above This can be recognized using almost the same processing as in the embodiment.

Note that the present invention is not limited to the above-mentioned embodiments, and for example, the third embodiment
The character inclination detection section 8 shown in the figure may be constructed of other functional blocks, or these functional blocks may be executed under program control using a computer.

Furthermore, the character slant detection processing is performed in Fig. 6 (a>, (b)
) Various modifications are possible, such as executing with a processing flowchart other than ).

(Effects of the Invention) As explained in detail above, according to the present invention, in the plane expressed by the X-Y coordinate system including the character pattern, αχ
The coordinates giving the maximum and minimum values of βy are determined, and the inclination angle of the character pattern is detected based on the coordinates. Next, a marginal distribution is extracted, and dividing points are detected using the marginal distribution. Thereafter, a dividing boundary within the character circumscribing frame is set based on the inclination angle and peripheral distribution, and features are extracted for each area divided by the dividing boundary. Therefore, the extracted features are stable even for character patterns having an inclination, and there is no need to prepare a dictionary that accommodates the deformation of the character inclination. As a result, the processing speed can be increased by reducing the dictionary capacity and the collation time thereby, and a character recognition method with high recognition accuracy can be realized on a small hardware scale.

[Brief explanation of drawings]

Fig. 1 shows an embodiment of the present invention, and shows a block diagram of a character recognition device for explaining a character recognition method, and Fig. 2 shows an embodiment of the present invention.
Figures (a) and (b) are diagrams for explaining the conventional character recognition method, Figure 3 is a functional block diagram of the character slant detection section in Figure 1, Figure 4 is a diagram showing an example of italic characters, Figure 5 (a>
, (b) is a diagram showing an example of detection of feature points to explain the slope detection method, Figures 6 (a) and (b) are flowcharts of character slope detection, and Figures 7 (a) and (b) + is a diagram for explaining a marginal distribution extraction method. 1...Photoelectric conversion unit, 2...Line buffer, 3...Character cutting unit, 4...Bun register, 5... Character frame detection section, 6...
Line width measurement section, 7... Sub-battle extraction section, 8...
...Character slope detection section, 9-----0: Marginal distribution extraction section, 10... Division point detection section, 11...
Feature matrix extraction unit, 12...Identification unit, 21
. . . X coordinate generation means, 22 . . . Y coordinate generation means, 30, 40 . . . Coordinate detection means, 50
...Inclination calculation means. Mochi rice letter certificate, san, method 2 figure 3? Examples of fonts Figure 4 Inclination Pr degree detection method Figure 5 Tai small inert history buried (b ん letter I ↑ (detection flow + - to cicada 6) Figure surrounding area 151 Yukokata Yudaisho Figure 7

Claims (1)

[Scope of Claims] A circumscribing frame of a character pattern obtained by photoelectrically converting characters on a medium is detected, the inside of the circumscribing frame is divided into a plurality of regions, and features are extracted for each divided region. Then, in the character recognition method that recognizes the character by comparing the extracted features with standard features prepared in advance, at least two sets of specified αx+βy (however, x,
Find the coordinates that give the maximum and minimum values (y is the coordinate value of the X-Y coordinate system), detect the inclination angle of the character pattern based on the coordinates, and scan the character pattern based on the inclination angle to determine the surrounding area. After extracting the distribution and detecting dividing points within the circumscribing frame using the marginal distribution, setting dividing boundaries within the circumscribing frame based on the slope angle and dividing points, and determining the dividing points within the circumscribing frame based on the dividing boundaries. A character recognition method characterized by dividing the inside of a frame into multiple regions.