JPH06274695A - Character recognition method - Google Patents

Abstract

PURPOSE:To obtain a character recognition method in which a dictionary can be easily generated by calculating the distribution of a chain code in the prescribed area with designated contour of a character image, and recognizing a character in detail based on the distribution. CONSTITUTION:A binarization processing part 1 binarizes image data, and a runlength data generating part 2 generates runlength data from binary image data, and a vertical phase expression data generating part 3 generates vertical phase expression. Also, a middle classification condition generating part 4 generates a middle classification condition by using the position of a run in the vertical phase expression. Furthermore, a chain code distribution generating part 5 generates chain code distribution normalized for a partial contour and normalized code length distribution, and delivers them to a character recognizing part 7, and the character recognition part 7 performs recognition processing based on the distribution. In such a way, since the dictionary can be generated from sampling data by using a statistical method, the dictionary can be easily generated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition method that facilitates dictionary creation.

[0002]

2. Description of the Related Art In a method for recognizing a binarized character image, the present applicant has previously generated run data for each row from binarized image data, and a run data pattern between adjacent rows can be obtained. Extract different pairs of rows (phase change pairs),
A method has been proposed in which a line figure is expressed by a series of a pair of lines (vertical phase expression), and a character is recognized using the vertical phase expression (Japanese Patent Application No. 4-46574). The method using this vertical phase representation removes noise on a very small amount of compressed data as compared to the conventional method that performs noise removal on the entire pixel (in the above-mentioned application, this noise removal is That is, there is an advantage that it is fast.

A character recognition method using the vertical phase representation will be briefly described. Binarized image data (FIG. 10)
Is represented by a vertical phase composed of the phase change pairs 21 to 24 (25 in FIG. 11).

A representation of the connection state information in the vertical phase representation by one character string is called a topology code in the vertical phase representation and is also called a "model". Classification is done by determining whether the model is the same as in the dictionary. FIG. 12 shows ten types of topology codes (including other codes *). For example, the number "2" is d, A, pI, p (see FIG. 1).
3). In the previously proposed character recognition method, the characters are roughly classified according to the combination of the topology codes of the vertical phase representation.

However, when recognizing a character as shown in FIG. 14, all the characters "6", "4" and "0" have the same topology code (d, Id, IA, VI, V,
p), it is not enough to use the topology code of the vertical phase expression. As described above, the topology code of the vertical phase representation can perform only the major classification of characters, and thus the classification needs to be performed in more detail.

As a character recognition method that solves such a point, the present applicant has already proposed a character recognition method using gi data and a character recognition method using a direction code (Japanese Patent Application No. 4-52213). . That is, the classification based on the gi data is a classification performed using the amount (numerical value) information easily obtained from the vertical phase expression, and the height information of the phase change pair included in the vertical phase expression and the vertical phase expression. The position information of the run indicated by is used. Then, the classification is performed by a simple logical expression in which the classification by the threshold value is combined.

For example, the topology code (d,
The characters that generate Id, IA, VI, V, p) are 98 characters, and the breakdown is 96 examples of the character "6" and the character "0".
And the letter "4" was one example each. Then, the model of FIG. 15 (d, Id, IA, VI, V, p)
In the above, each phase change pair is gi0, gi in order from the top.
1, ... gi5, the height difference between gi5 and gi4 is a, the height difference between gi1 and gi0 is b, the height difference between gi4 and gi2 is c, and the height difference between gi5 and gi0 is If h, then the above model (d, Id, IA, VI, V, p)
The data of 98 characters belonging to could be completely separated by the following conditions using a, b, c and h.

Condition for character "6": a / h <0.15 and 0.25≤b / h <0.80 Condition for character "0": a / h <0.15 and b / h <0.
Condition of 15 characters “4”: 0.25 ≦ a / h <0.60.

Further, the classification by the direction code enables the classification of numbers which cannot be classified by the method using the gi data, and the direction code (FIG. 16) is sampled at equal intervals for a part of the contour. The points are taken and the slopes of the points are coded and arranged. Then, the detailed classification is performed using an automaton for character recognition.

[0010]

However, in the above detailed classification method, it is necessary to consider the structure of the automaton while referring to the data, and there is a problem that a great amount of labor is required to create the automaton. In addition, the above-mentioned method has a drawback that training is required to create an automaton because the automaton is far from the human intuition.

An object of the present invention is to provide a character recognition method that facilitates dictionary creation.

[0012]

In order to achieve the above object, in the invention according to claim 1, run data is generated for each row from binarized image data, and a run data pattern between adjacent rows is generated. In a method of extracting a different phase change pair and recognizing a character by using a vertical phase expression composed of connected states of runs in the phase change pair, a part or all of the outline of a character image is specified and the specified It is characterized in that the distribution of the chain code in a predetermined area of the contour is calculated and the characters are recognized in detail based on the distribution.

According to the second aspect of the invention, run data is generated for each row from the binarized image data, and phase change pairs having different run data patterns between adjacent rows are extracted.
In a method for recognizing a character by using a vertical phase expression composed of connected states of runs in the phase change pair, a part or all of the outline of a character image is specified, and the specified outline is approximated by a polygon. However, the feature is that the character is recognized in detail by using the expression.

According to the third aspect of the invention, run data is generated for each row from the binarized image data, and phase change pairs having different run data patterns between adjacent rows are extracted.
In a method for recognizing a character by using a vertical phase expression composed of connected states of runs in the phase change pair, a part or all of the outline of a character image is specified, and the specified outline is expressed as a collection of symbols. However, the feature is that the character is recognized in detail by using the expression.

According to a fourth aspect of the present invention, run data is generated for each row from the binarized image data, and phase change pairs having different run data patterns between adjacent rows are extracted.
In a method for recognizing a character by using a vertical phase representation composed of connected states of runs in the phase change pair, a part or all of the outline of a character image is specified, the specified outline is Fourier-transformed, and The feature is that the character is recognized in detail based on the Fourier-transformed feature.

[0016]

The binarization processing unit binarizes the image data, the run length data generation unit creates run length data from the binarized image data, and the vertical phase expression data generation unit creates vertical phase expression. To do. The middle classification condition generation unit generates the middle classification condition using the position of the run in the vertical phase expression. The chain code distribution generation unit generates a normalized chain code distribution and a normalized code length distribution for the partial contour, and passes them to the character recognition unit. The character recognition unit performs recognition processing based on the distribution.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be specifically described below with reference to the drawings. <Method Using Chain Code> FIG. 1 shows a first method of the present invention.
It is a block diagram of an embodiment of the. In the figure, 1 is
A binarization processing unit that reads an image and binarizes it, a run length data generation unit that generates run length data from the binarized image data, and a phase change from run length data of adjacent rows. A vertical phase expression data generation unit that extracts a pair and generates a vertical phase expression. The binarization processing unit 1, the run length data generation unit 2, and the vertical phase expression data generation unit 3 are described in the above-mentioned application (Japanese Patent Application No. 4-46574).
No.) has the same function and configuration as those described in No.

Reference numeral 4 denotes a middle classification condition generating section for generating middle classification conditions by using run positions in the vertical phase expression, and uses simple arithmetic expressions and threshold values while referring to the vertical phase expression data and the run length data. To generate middle classification conditions. The middle classification condition generating unit 4 is the same as the above-mentioned application (Japanese Patent Application No. 4-52).
No. 213) has the same function and configuration (however, in the present invention, the gi data classification unit of the previous application is referred to as a middle classification condition generation unit).

Reference numeral 5 is a chain code distribution generator for obtaining the distribution of chain codes for the partial contour. 6 is
It is a noise removing unit that performs one shaving process by referring to the vertical phase expression data, and changes the vertical phase expression data and the run length data. This noise removing unit has the same function and configuration as those described in the above-mentioned application (Japanese Patent Application No. 4-46574). A character recognition unit 7 performs detailed recognition based on the vertical phase expression, the middle classification condition, and the chain code distribution.

The character recognition method of the present invention will be described in detail with reference to the image data shown in FIG. As shown in FIG.
From the image data binarized by the binarization processing unit 1,
The run length data generation unit 2 creates run length data, and the vertical phase representation data generation unit 3 creates vertical phase representation from this run length data as described above.
Vertical phase representation of image data in Figure 2 (topology code)
Becomes "d, A, V, p".

The middle classification condition generating unit 4 generates middle classification conditions as shown in FIG. That is, the model “d, A, V,
When x = a / b for p ″, x <0.85
The data is classified by (group 1), 0.85 ≦ x <1.25 (group 2), 1.25 ≦ x (group 3). The image data shown in FIG. 2 is “d, A, V, p”.
It is classified into Group 1 of.

The chain code distribution generator 5 is shown in FIG.
As shown in (a), a chain code is generated for the classified “d, A, V, p” partial contours (r, r ′) of group 1. FIGS. 6A and 6B are diagrams for explaining the definition of the chain code, and as shown in FIG.
The code of the direction is determined, and the outline of the figure is expressed by the chain of the code as shown in FIG. 6 (b).

The chain code distribution generation unit 5 generates a chain code distribution for the partial contour designated by the character recognition unit 7 as shown below. Specifically, in the case of the image data of FIG. 2, the character recognition unit 7 instructs the chain code distribution generation unit 5 to display “partial contour r of FIG.
Is divided into ⅓, ⅓ and ⅓, and “partial contour r ′ in FIG.
4, 13/20, 1/10, and a chain code distribution ”is given.

In the following, the partial contour r of FIG.
An example of obtaining a chain code distribution by dividing into 1/3 and 1/3 will be described with reference to the flowchart of FIG. In step 500, the partial contour and the division ratio (1/3,
1/3, 1/3) is given. In step 501,
The given contour is divided based on the division ratio. Hereinafter, the divided contour is referred to as a divided contour. FIG. 4B shows an example in which the partial contour r is divided into 1: 1: 1, and the respective divided portions are designated as r1, r2, and r3.

In step 502, the first divided contour r1 becomes R. In step 503, a chain code is generated for R. In the case of this example, “54544
A chain code "554555556565556" is generated. In step 504, the distribution and length of the chain code obtained in step 503 are obtained. In this case, the distribution is “4 times for code 4, 12 times for code 5, 3 times for code 6”, and the length is 19 characters.

In step 505, the chain code distribution obtained in step 504 is normalized. In the case of this example, “0 is 0%, 1 is 0%, 2 is 0%, 3 is 0%, 4
Is 21%, 5 is 63%, 6 is 16%, and 7 is 0%. " After that, this is "0,0,0,0,21,63,1
6, 0 ".

In step 506, the normalized chain code distribution is output. In step 507, it is determined whether the processing has been completed for all the divided contours. In this case, since the divided contours r2 and r3 have not been processed yet, the process goes to step 502. When this loop is repeated two more times, “566666607” for the divided contour r2.
After obtaining the chain code of length 26 of 707700010107656665 "," 27,8,0,0,
The normalized chain code "0, 12, 34, 19" is output, and "665665" is output for the divided contour r3.
65665667 "After the length 14 chain code is obtained, the normalized chain code" 0,0,0,0,0,29,64,7 "is output, and step 508 is performed.
go to.

In step 508, the length of each divided contour is 1
9 characters, 26 characters and 14 characters are normalized to "r1, r
The results are 32%, 44%, and 24% for 2 and r3, respectively. In step 509, the result "32, 44, 24" is output.

The normalized chain code distribution "0,0,0,0,21,63,1" obtained as described above
6,0 "," 27,8,0,0,0,12,34,1
The character recognition unit 7 performs character recognition based on "9", "0,0,0,0,0,29,64,7" and the normalized code length distribution "32,44,24". The character recognition unit 7 uses a dictionary in which the average and standard deviation of each element have been obtained in advance from sample data, and performs recognition according to how far each distribution value is from the average. The average and standard deviation required for the dictionary can be calculated automatically.

In the above example, if "the group 1 of" d, A, V, p "is", the partial contour r of FIG.
The rule "obtain the chain code distribution by dividing it into 3, 1/3, 1/3" has to be created by a human, but the labor is significantly less than that in creating an automaton.

<Method Using Polygonal Approximation> FIG. 7 is a block diagram of the second embodiment of the present invention. In FIG. 7, the binarization processing unit 1, the run length data generation unit 2, the vertical phase expression data generation unit 3, the middle classification condition generation unit 4, and the noise removal unit 6 are the same as those in the first embodiment described above. . 5
Is a polygon expression generation unit for obtaining a polygon expression for the partial contour, and 7 is a character recognition unit for performing detailed recognition based on the vertical phase expression, the middle classification condition, and the polygon expression.

The character recognition method of the second embodiment will be described by taking the image data shown in FIG. 2 as an example. In the same manner as described above, the image data shown in FIG.
The polygonal representation is generated for the partial contours r and r ′ of FIG.

The polygonal expression generating unit 5 generates a polygonal approximation of the contour specified by the character recognition unit 7. For the generation of the approximate expression of the polygon, for example, methods such as "Williams method", "Ramer method", "L1 approximation", and "division fusion method" are used. Others, Basics of Image Recognition [1], Ohmsha, 1986, pp173-
190). The character recognition unit 7 performs recognition processing using the polygonal expression thus obtained.

<Method Using Symbols> FIG. 8 is a block diagram of the third embodiment of the present invention. In FIG. 8, the binarization processing unit 1, the run length data generation unit 2, the vertical phase expression data generation unit 3, the middle classification condition generation unit 4, and the noise removal unit 6 are the same as those of the above-described first embodiment. . 5
Is a symbol expression generation unit that obtains a symbol expression for the partial contour, and 7 is a character recognition unit that performs detailed recognition based on the vertical phase expression, the middle classification condition, and the symbol expression.

The character recognition method of the third embodiment will be described by taking the image data shown in FIG. 2 as an example. In the same manner as described above, the image data shown in FIG.
The symbol representation is generated for the partial contours r and r ′ of FIG.

In the symbol expression generation unit 5, the character recognition unit 7
Generates a symbolic representation for the contour specified by. For the generation of this symbol expression, the method proposed by the present applicant to express an outline by a group of straight lines having an overlap (Japanese Patent Application No. 3-253185) is used. The character recognition unit 7 performs recognition processing using the symbolic expression thus obtained.

<Method Using Fourier Transform> FIG. 9 is a block diagram of the fourth embodiment of the present invention. In FIG. 9, a binarization processing unit 1, a run length data generation unit 2,
The vertical phase expression data generation unit 3, the middle classification condition generation unit 4, and the noise removal unit 6 are the same as those in the first embodiment described above.
Reference numeral 5 is a Fourier transform expression generation unit that obtains a Fourier transform expression for a partial contour, 7 is a vertical phase expression, a middle classification condition,
And a character recognition unit for performing detailed recognition based on Fourier transform representation.

The character recognition method of the fourth embodiment will be described by taking the image data shown in FIG. 2 as an example. In the same manner as described above, the image data shown in FIG.
It is classified into the group 1 of p ″, and the Fourier transform expression is generated for the partial contours r and r ′ of FIG.

That is, the Fourier transform representation generating section 5 generates a feature in the spatial frequency domain by performing a Fourier transform on the contour designated by the character recognizing section 7. As a method of generating this Fourier transform expression, an FFT (Fast Fourier Transform) is used.
orm). The character recognition unit 7 performs recognition processing based on the Fourier-transformed features.

It should be noted that the present invention is not limited to the above-described ones, but may be modified so that, for example, the above-mentioned methods are combined, or features are obtained by using an automaton, and recognition is performed based on the obtained features. It is possible.

[0041]

As described above, according to the first and fourth aspects of the present invention, the dictionary can be created from the sample data by using the statistical method, so that the dictionary can be created easily.

According to the second and third aspects of the present invention, the dictionary can be created using polygons and symbolic expressions that match the intuition of a person, so that the dictionary can be created easily.

[Brief description of drawings]

FIG. 1 is a block configuration diagram of a first embodiment of the present invention.

FIG. 2 is an example of binarized image data.

FIG. 3 is a diagram illustrating medium classification conditions of a model “d, A, V, p”.

4A is a diagram showing a partial contour, and FIG. 4B is a diagram showing an example in which the partial contour is divided into divided contours.

FIG. 5 is a processing flowchart of a chain code distribution generation unit.

6A and 6B are diagrams for explaining the definition of a chain code.

FIG. 7 is a block configuration diagram of a second embodiment of the present invention.

FIG. 8 is a block configuration diagram of a third embodiment of the present invention.

FIG. 9 is a block configuration diagram of a fourth embodiment of the present invention.

FIG. 10 is binarized image data.

11 is a vertical phase representation diagram of FIG. 10. FIG.

FIG. 12 is a diagram showing a topology code.

FIG. 13 is a diagram showing a vertical phase representation of a numeral 2 and a topology code.

FIG. 14 is a diagram showing an example in which characters cannot be classified only by a topology code.

FIG. 15 is a diagram illustrating classification by gi data.

FIG. 16 is a diagram showing a direction code.

[Explanation of symbols]

 1 Binarization processing unit 2 Run length data generation unit 3 Vertical phase expression data generation unit 4 Medium classification condition generation unit 5 Chain code distribution generation unit 6 Noise removal unit 7 Character recognition unit

Claims (4)

[Claims] 1. Run data is generated for each row from binarized image data, phase change pairs with different run data patterns between adjacent rows are extracted, and the run connection state in the phase change pair is configured. In the method for recognizing a character using the vertical phase representation, a part or all of the outline of the character image is specified, the chain code distribution in a predetermined area of the specified outline is calculated, and the character is calculated based on the distribution. A character recognition method characterized by recognizing in detail. 2. Run data is generated for each row from binarized image data, phase change pairs with different run data patterns between adjacent rows are extracted, and the run connection state in the phase change pair is configured. In a method for recognizing a character using the vertical phase expression, a part or all of the outline of the character image is specified, and the specified outline is approximated by a polygon,
A character recognition method characterized in that a character is recognized in detail using the expression. 3. Run data is generated for each row from binarized image data, phase change pairs having different run data patterns between adjacent rows are extracted, and the run connection state in the phase change pair is configured. In a method for recognizing a character using a vertical phase expression, a part or all of the outline of a character image is specified, the specified outline is expressed as a set of symbols, and the character is recognized in detail using the expression. A character recognition method characterized by the above. 4. Run data is generated for each row from binarized image data, phase change pairs having different run data patterns between adjacent rows are extracted, and the run connection state in the phase change pair is configured. In a method for recognizing a character using a vertical phase representation, a part or all of the outline of a character image is specified, the specified outline is Fourier-transformed, and the character is recognized in detail based on the Fourier-transformed feature. A character recognition method characterized by: