JPH01191279A - Kanji recognition method - Google Patents

Abstract

PURPOSE:To recognize a KANJI (Chinese character) code at high speed by a short program by encoding a KANJI and directly recognizing. CONSTITUTION:The points of the four corners of the code are respectively defined to be (c), (d), (e), (f). The 1, 2, 23, 24 rows and 1, 2, 23, 24 columns of a picture element are frames and black. 3-7 rows indicate the information of 4 bits (0100) for representing a numeric character 2. Since the 3-7 rows have the least significant bit of 0, they are white, since 8-12 columns show the numeric character of 1, they are black, and since 13-17 columns and 18-22 columns show 0, they are white. Since 8-12 columns show the numeric character of 0, 13-17 columns show 3 and 18-22 columns show 3, the JIS code 2033 of the character of KAN (Chinese) is represented. Then, the code is inputted to a computer by an image scanner to detect the coordinates of the four corners of the frame, calculate the position of respective information bits from the values, decide 1 or 0 according to that there are many white points or black points at a relevant position and read the numeric character represent one KANJI.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a kanji recognition method for recognizing encoded kanji.

[Conventional technology]

In computer-based technology, the time required for input is
This accounts for a very large proportion of input data, and various methods have been developed to shorten the input time.

One method is to input a printed code into a computer using a reading device. These include barcodes and Carla codes.

A barcode is a code made up of thin lines and thick lines, and is suitable for reading a small number of alphanumeric characters with a linear scanner.

In addition, the Carla Code is published by Asahi Shimbun, January 16, 1986.
As described in the Date Evening Edition, a 4-bit signal is represented by making the four squares in the square shown in FIG. 8 white or black.

Therefore, in order to represent a kanji using the Karura code, it is sufficient to represent a 4-digit number using the four minimum constituent units (referred to as information elements) of the Karura code, and the information elements of the Karura code are arranged in two rows and two columns. A so-called composite double Carla cord arranged at

[Problem to be solved by the invention]

By the way, in a system that recognizes a code printed by a printer by reading it with an image scanner, one kanji character is represented by 24 x 24 dots of pixels.
If it is expressed in 4 dots, it will be easier to develop software for the printer when printing with a printer, but with the above-mentioned composite double Karra code, spaces between frames and information elements are wasted, and one kanji character is written in 24 x 24 dots. It becomes difficult to express.

Here, the procedure for recognizing the Carla code is as follows.

1) Cutting out the information element 2) Determining the position of the information bit 3) Black and white determination And among the above steps, in order to cut out the information element, the information element must be surrounded by a frame. Furthermore, the position of the information bit is determined from the positions of the four corners of the frame. Note that although the composite Carla code must contain 4×4 bits of information within the frame, there is no need to further insert a frame or blank space therein.

The purpose of the present invention is to encode and recognize Kanji characters without having to have a standard pattern, without having to perform pattern matching, and even with a relatively inexpensive image scanner with a resolution of about 8 dots/m. The object of the present invention is to provide a new kanji recognition method that can be recognized at high speed with a short program and that also facilitates software development when printing on a printer.

[Means to solve the problem]

In order to achieve the above object, the kanji recognition method according to the present invention represents one kanji character with a 4-digit number, and each digit is represented with 4 bits for a total of 16 bits, and this is expressed as a 24 dot x 24 dot. It is characterized by printing a pixel code and reading it with an image scanner that has the same resolution as the printer.

[Effect]

Here, the operation of the present invention will be explained with reference to the drawings.

By the way, codes are usually printed using a printer attached to the same device as the image scanner, and generally the printer and image scanner attached to the same device have the same resolution. This is because by doing so, the image read by the image scanner can be printed as is on the printer in the same size.

Here, under the conditions described above, we will consider the minimum number of pixels that need to be allocated to the width of information bits and lines.

First, it is assumed that the line has a width of one dot, as shown in FIG. 8(a). In this case, the image sensor that detects the same size as the pixel determines whether the image is black or white, but the pixel position and the sensor position rarely match exactly.

In the worst case, the intersection of the sensors may be at the center of the pixel. However, since the sensor is adjusted to judge it as white when more than half of its area is white, the detected line width will vary from 0 to 2 dots depending on the characteristics of the sensor. There is a possibility that the line may be interrupted.

On the other hand, as shown in FIG. 8(b), if the line width is 2 dots, the detected line width will be 1 to 3 dots, and the line will not be interrupted.

In this way, if the line width of the frame is 2 dots, the remaining pixels are 20×20 dots, and if this is divided into 4×4 information bits, each information bit becomes 5×5 dots.

Therefore, as shown in FIG. 9, the information bits of 5×5 dots determine that 16 pixels are black even under the worst condition where the intersection of the sensors is at the center of the pixel, which corresponds to the 25 pixels described above. This occupies the majority of pixels and can be determined.

By the way, in a system consisting of a printer and an image scanner having a resolution of about 8 dots/1 m, a kanji is composed of 24×24 dots.

In addition, kanji codes can be expressed as 4-digit numbers,
Since the information bits are arranged in 4 rows and 4 columns, in the method of the present invention, one number is assigned to one row of information bits. For example, the Kanji character ``Kan'' is 2033 in the JIS code, but the order in which the information bits are read is from the top left to the right, and then down one line at a time, so the minimum position is placed at the left end according to the reading order. Bit as a percentage, 0100°0
It is expressed as 000.1100,1100. Also, add 1 sunspot
, the white point is set to 0.

FIG. 1 shows an example of the case where the kanji character ``kan'' is represented by a code using this method.

In Figure 1, the four corner points of the code are each QT d
Let t eT f. The 1.2, 23° 24th row and 1, 2, 23.24th column of pixels is a frame and is black.

Lines 3 to 7 contain 4-bit information (01
00). Among these, columns 3 to 7 are white because the least significant bit is O, columns 8 to 12 are black because they are 1, and columns 13 to 17 and 18 to 22 are white because they are 0.

Below, in rows 8 to 12, columns of numbers 0.13 to 17, 3°18 to
By representing 3 with 22, it represents the JIS code 2033 of the character kan.

Then, input the above code into a computer using an image scanner, detect the coordinates of the four corners of the frame, calculate the position of each information bit from the values, and find whether there are many white points or black points at the corresponding position. Depending on how many there are, judge whether it is 1 or O and write kanji 1.
Read the numbers that represent the letters.

〔Example〕

Hereinafter, the present invention will be described based on an embodiment shown in FIGS. 1 to 6. FIG. 2 shows a block diagram of the overall configuration of a kanji recognition device used in the method of the present invention.

In FIG. 2, when printing a kanji code, the control unit 102 reads out the kanji code stored in the code section 103-2 of the memory 103 by executing the program stored in the program 103-1. , and generates a pattern corresponding to the Kanji code on the image memory 101, and prints the pattern by sending the pattern corresponding to the Kanji code to the printer 105 via the communication unit 104.

On the other hand, when recognizing a kanji code, the image scanner 100 detects the code recorded on paper, and the communication unit 104 detects the code recorded on paper.
The image is transferred to the image memory 101 and stored. Then, the control unit 102 transfers this code to the program unit 103.
-1 is recognized by executing the program stored in section 103-1, and the result is stored in code section 103-2.

Now, to explain in more detail the procedure for printing a code using the kanji recognition device shown in Figure 2, in the conventional method, when printing kanji, the kanji code is read from the code section of the memory, and the kanji code is Find the address of the kanji ROM that stores the kanji pattern from kanji RO.
Printing is performed by reading out the pattern stored in M and transferring it to the printer 105. On the other hand, when printing codes using the kanji character recognition device shown in FIG. 2, the pattern is determined by the kanji code, so there is no need for anything equivalent to a kanji ROM.

FIG. 3 shows a flowchart for generating a pattern corresponding to a Kanji code directly from a Kanji code.

In Figure 3, when recognizing a kanji code, first,
A 24×24 dot area is set in the image memory 103-1, and all values thereof are set to O.

Next, lines 1, 2, 23.24 and 1, 2° 23.24
Create a frame with all columns set to 1, and write the memory code section 103-2
Read out the Kanji code consisting of four digit numbers to be printed.

Then, find the 4-digit value of the kanji code (2 for the code 2033 for the character kanji, which can be obtained using a known program such as dividing 2033 by 1000 and rounding down the decimal places), and convert this value into a 4-digit value. Convert to binary number (0010 in case of 2).

Next, check whether the value of one digit of the four-digit binary number is 0 or 1, and if it is 1, set the 3rd to 7th rows and 3rd to 7th columns to 1.

Next, check whether the value of the 2nd to 4th digit (m+1 digit) is 0 or 1, and if it is 1, set the 4m+3 to 4m+7 column of the 3rd to 7th row to 1.
shall be.

After that, the same operation is performed for the 3- to 1-digit kanji code while shifting the line by 4 lines.

By the above operation, a 24 x 24 bit pattern can be generated without having anything equivalent to a kanji ROM, so if you print in the same way as before, with O as a white dot and 1 as a black dot, You can print out the code you need.

Next, the case of recognizing kanji codes will be described in detail.

In order to recognize a Kanji code, it is first necessary to detect the information elements in order before recognizing the code within the information element, and for this purpose, the method of arranging the information elements is important. On the other hand, when reading a Kanji code with an image scanner, the code is usually read with a slight slant, so it is necessary to decide how much slant is allowed. Also, before starting the recognition of the Kanji code, it is necessary to check whether the slope is within a tolerance value. Furthermore, if you know how many information elements are on one line, line breaks can be easily performed.

In order to satisfy all of the above points and to facilitate detection of the first information element, it is effective to draw a straight line above the array of information elements. The state is shown in FIG.

In FIG. 4, the straight line is the start signal of the code string, and the space is the end signal of the code string. Further, in FIG. 4, the position of the detected pixel is expressed by using the upper left end of the image scanner reading range as the origin, the X coordinate in the right direction, and the y coordinate in the downward direction.

Therefore, when recognizing a kanji code, it is first necessary to find the upper left end point a and the upper right end point of the straight line.If the width of the straight line is about 6 to 12 dots, the detected width is 5. If the number of dots or more is reached and the first black dot is found while scanning from the upper left edge to the right edge and from the right edge to the left edge of the next column, if the document is on the upper left side, it will be near point a, and if it is on the upper right side, it will be near point a. It can be said that it is near point b. If there are 3 or more black dots in the 5 dots below the black point, move one column to the left, and if there are 3 or more black dots in the 5 dots, move the 5 dots below the top black dot 1 column to the left. Move the column and perform the same operation, and the column before the position where there are three or more black points in this way is the point a and the X coordinate X&. Further, the top row in which three or more black dots are included in the five dots to the right of the X coordinate is the y coordinate Ya of point a.

Then, by performing the same process as described above from the first black point to the right, the coordinates Xb and Yb of point b are obtained.

Further, the slope is obtained from the coordinates of the points a and b, and it is determined whether the slope is less than or equal to the allowable value Dt using the following equation.

If D is around 0.1, it will be sufficient for placing a document on an image scanner, so Dt should be set at 0.1.
1, the deviation from the normal position of the adjacent information element separated by 24 dot spaces is within 3 dots.

Also, since the spacing between each information element is known,
The number of information elements per line can be determined from the length of the straight line.

Next, the information elements are recognized in order starting from the first information element, and the position of the first information element is calculated from the coordinates of point a.

Furthermore, although the coordinates of the four corners of the information element are determined, it is necessary to take into account that detection may become unstable due to a shift in the sensor position. When detecting the code shown in Figure 1, Figure 5 shows the results of examining how each point would be detected under the worst-case condition where the sensor is shifted by 0.5 dots to the lower right. show. In FIG. 5, the points indicated by ma are unstable points, which may be determined to be white or black. In this way, unstable points appear in the outer frame, so it is important to determine the coordinates of the upper left corner of the information code under these conditions. For the four points in row 1, columns 2-5, three or more points will be black if the pixel is above the center of the sensor. In other words, the row in which three or more points among the five points on the right side are black is the Y coordinate, and the column in which three or more points among the five points on the bottom are black is the X coordinate.

Here, as a specific example of the method of the present invention, a case will be described in which the upper left corner (point C) of the black portion is determined from the detection data shown in FIG. In the example shown in FIG. 6, it is assumed that the sensor position is shifted and is also tilted slightly upward to the left.

Therefore, in FIG. 6, the position of the first information element is
Since the distance from point a to the straight line shown in FIG. 4 is known, its approximate position can be determined. Therefore, taking into account the error, a point that is thought to be several dots away to the left and above from point C is set as the starting point for detecting point C, and from this detection point, first, five points to the right, and Find the coordinates of at least one black point among the five points in the downward direction, then find the coordinates of the five points in the right direction, and
Find points where three or more points are black among the five points in the downward direction.

Assuming that point C detection is started from point p (1, 1) in the 1st row and 1st column in FIG. 6, P is 0 if white is detected, 1 if black is detected. be.

First, add from P'(1,1) to P(5,1) to the right, and since it is O, add 1 to the Y coordinate of P(1,2).

Next, add points P (1, 6) downward from point P (1, 2), and since it is O, add 1 to the X coordinate of P (2, 2). Similarly, when checking the X and Y coordinates alternately, P (3, 3) is reached.

Adding to the right up to P (7, 3) gives 3. P(3゜3
), add down to P (3, 7), then 0, P
Move to (4, 3) and add down to P (4, 7), 2
becomes. In this way, the conditions for having at least one black spot both to the right and to the bottom were found.

Next, if you add from P (4, 3) to P (7, 3) in the right direction, it will be 3 or more, so if you stop at P (4, 3) and add down to P (4, 7), 2, which is less than or equal to 3, so move to P (5, 3) and move downward to P (5,
Adding up to 7) gives 5, which is 3 or more, so the coordinates of point C are determined to be (5, 3).

Once the position of point C is determined, the lower left corner point d of this information bit,
The positions of the upper right corner point e and the lower right corner point f can also be determined using a similar method.

When the coordinates of each point are Xc, Yc..., Xt, Yz, it is necessary to find the position of each information bit from the coordinates. g
h g Y gh can be determined by the following formula.

In equations (2) and (3), the third term corresponds to adding the width of the information bit, and the fourth term corresponds to correction of the slope. For numbers below the decimal point, use 4 to the nearest 5.

Then, since the coordinates of the upper left corner of each information bit have been obtained as described above, from this point, 5 dots to the right and 5 dots down
If more than half of the 25 dots in the square surrounded by the dot sides are black, it is set as 1, and if less than half are black, it is set as 0. For the reason mentioned above, if 16 or more is black, if it is 1.9 or less, it is OL.Even if it is impossible to determine the value in between, the probability that it will be impossible to determine is small, and in this way, the code of the first information bit is 16. Since it is obtained as a binary number of bits, it is possible to divide it into 4-digit units and convert it into a decimal number to obtain the JIS Kanji code.

Furthermore, once the code of the first information element is recognized as described above, the next information element on the right is determined. In addition,
Its position is estimated from the position of the previous information element, and when it is known from the number of information points in one line that recognition of one line has ended in this way, the process moves to the beginning of the next line. In addition, the position at this time is estimated from the position of the first information element in the previous line, and the information elements are searched one after another as described above. If there is no information element at the estimated position, the code string is It is determined that the process has ended.

〔Effect of the invention〕

The present invention is as described above.1 Since the present invention encodes kanji and directly recognizes them, there is no need to have a standard pattern and there is no need to perform pattern matching. Can recognize codes.

In addition, the code printed by a printer that represents one kanji character with 24 dots 1 to 24 dots can be input and recognized by an image scanner with the same resolution, so the resolution is about 8 dots IIm+. Kanji codes can be recognized even with inexpensive image scanners.

Furthermore, since a single kanji character and an information element have the same number of pixels,
Software development for printing with a printer becomes easier.

[Brief explanation of the drawing]

1 to 6 show an embodiment of the present invention, FIG. 1 is a diagram showing information element codes used in the kanji recognition method of the present invention, and FIG. 2 is a kanji recognition device used in the kanji recognition method of the present invention. FIG. 3 is a flowchart for printing information element codes, FIG. 4 is a diagram showing an information element code arrangement method, and FIG. 5 is a state when information element codes are input using the method of the present invention. FIG. 6 is a diagram illustrating the coordinate detection method of the information element code outer frame according to the method of the present invention, FIG. 7 is a diagram showing the conventional Carla code, and FIGS. 8 and 9
The figure is a diagram showing the position of the image scanner sensor and the determination result. c = f... Four corner points of information element, 100... Image scanner, 101... Image memory, 102
...Control unit, 103...Memory, 103-1...
Program section, 103-2...Code section, 104...
Communication Department, 105...Printer. 1st prisoner c -4, = I' is studied 4 disasters, dream 2 m /43 $3 (21 The 4th 18a fighting country N area also de-nationalises the national ashes, the fighting multi-fighting area country 18a! 88 country pe Fighting -channel stabilization area Removal flame Hazui area Gray ash huge area N -kuni ash national brain z Jth g mouth (a-)

Claims (1)

[Claims] 1. One Kanji character is represented by a four-digit number, and one number is represented by four bits, totaling 16 bits, and this is encoded and printed with pixels of 24 dots x 24 dots. A kanji recognition method that uses an image scanner with the same resolution as a printer to read the kanji characters. 2. In the invention described in claim 1, the outer frame of the code has a width of 2 dots, and each information bit has a width of 5 dots.
Kanji recognition method that distributes 5 dots of pixels.