JP3301467B2 - Image pattern identification and recognition method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for identifying and recognizing a binary pattern in which an input pattern is expressed in white or black. Printed kanji, which is a typical example of a binary pattern,
When targeting many character and figure categories such as handwritten kanji, alphanumeric characters, symbols, mathematical formulas, figures, etc. Further, the present invention can be applied to a character recognition method for recognizing white characters and the like. Also, the present invention can be applied to a method of recognizing a plurality of categories of disorders and illnesses occurring in the system based on the results of diagnosis of various items used in disasters, medical treatment, and the like.

[0002]

2. Description of the Related Art Conventionally, kanji OCR (Optical Characte
r Reader) and character recognition processing devices and drawing recognition devices extract features for recognition from characters and graphic patterns in the form of vectors and compare them with standard pattern vectors of each category in a standard dictionary created in advance. A method is known in which a discriminant function such as a similarity or a difference is obtained between the characters and a category of the most similar character or figure is determined as a recognition result.
Conventionally, features that reflect the character line structure, such as character line direction, connection relationship, and positional relationship, are widely used as elements of feature vectors such as character recognition.However, designs with unique textures For images in which characters, blurring, crushing, character background noise, and the like are intense, these feature amounts greatly fluctuate, making it almost impossible to obtain sufficient recognition accuracy.

[0003] In addition, a black-and-white pattern input by conventional character recognition is used as an element of a feature vector, and a standard pattern vector of each category in a standard dictionary created in advance (expressed as a binary pattern like a feature vector) is used. Simple similarity is known as a discriminant function based on the number of black pixels between them. However, this method has a problem that an image composed of peculiar textures, such as design characters or character background noise, is erroneously recognized as a category having a large number of black pixels in the standard pattern vector.

[0004]

As described above, the conventional OCR and other recognition techniques correctly recognize characters that are erroneously recognized due to design characters, blurring, crushing, character background noise, and mixed white characters. There was a problem that a recognizable method was not sufficiently established. SUMMARY OF THE INVENTION An object of the present invention is to provide an image pattern identification / recognition method having a recognition function capable of coping with severe character deformation such as design characters, blurring, and crushing, and character strings in which white characters are mixed.

[0005]

[MEANS FOR SOLVING THE PROBLEMS] To achieve the above object
The invention according to claim 1, wherein the input binary image pattern
For each of the previously created binary categories for each recognition category,
The pattern to which the binary image pattern belongs.
Image category identification and recognition method
Thus, the first step of counting the black portions of the input binary image pattern
Step and the black portion of the input binary image pattern
Included in the black part of the binary mask pattern for each recognition category
A second step of counting the number, and the second step
The number counted (hereinafter referred to as Ci (i is the number of categories))
The number of black parts for each recognition category (hereinafter referred to as Bi (i is
A third step of comparing the number of categories)),
Ratio of the number counted in one step (hereinafter referred to as A) to Ci
A fourth step of compare, C in the third step
Recognition where i ≦ Bi and A = Ci in the fourth step
If there is only one category i, enter
Assume that the binary image pattern belongs to the recognition category i
It has an identification procedure .

[0006] The invention described in claim 2 is the invention according to claim 1.
In the image pattern identification / recognition method described in the above,
In the step, Ci ≦ Bi and in the fourth step, A =
When there are a plurality of recognition categories i serving as Ci,
Have a rejection procedure to reject the input binary image pattern
It is characterized by.

[0007] Further , the invention according to claim 3 provides an input
A recognition category created in advance for the binary image pattern
The binary image pattern is compared with each binary mask pattern for each
The image pattern that determines the recognition category to which the turn belongs
A method of determining and recognizing a white image of the input binary image pattern.
A first step of counting the number of copies, the input binary image pattern
Is the binary mask pattern for each recognition category
A second step of counting the number included in the white part of the
The number counted in the second step (hereinafter referred to as Ci (i is
Number of categories)) and the number of white parts for each recognition category (hereinafter B
i (where i is the number of categories)
And the number counted in the first step (hereinafter A and A)
A fourth step of comparing Ci) with Ci.
In the third step, Ci ≦ Bi and in the fourth step, A
= Ci, there is only one recognition category i
If the input binary image pattern is in the recognition category
It has an identification procedure that belongs to i.

[0008] The invention described in claim 4 is the invention according to claim 3.
In the image pattern identification / recognition method described in the above,
In the step, Ci ≦ Bi and in the fourth step, A =
When there are a plurality of recognition categories i serving as Ci,
Have a rejection procedure to reject the input binary image pattern
It is characterized by.

[0009]

[0010]

According to the above arrangement, the input binary image pattern
And a binary mask pattern for each recognition category that has been created in advance.
By checking the turn, the recognition pattern to which the binary image pattern belongs
Efficiently identify, recognize or reject categories
You.

[0011]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a configuration example of an image pattern identification / recognition processing apparatus to which the method of the present invention is applied. 1 is a binary image pattern storage circuit, 2 is a normalization processing circuit, and 3 is a binary mask pattern storage circuit. 4 is an identification / recognition determination circuit, 5 is a binary mask pattern creation circuit, and 6 is an image input / output device. The binary image pattern storage circuit 1 stores a binary image pattern of a character or a figure to be identified or recognized. The binary image pattern is composed of n pixels, for example, white pixels are “0” and black pixels are “1”.

The normalization processing circuit 2 inputs a binary image pattern composed of n pixels extracted from the binary image pattern storage circuit 1 and, for example, calculates a conventionally known centroid and 2
The position and size are normalized using the next moment. The binary mask pattern storage circuit 3 stores in advance one or more (generally a plurality) binary mask patterns of each category used for identification and recognition determination of an input image pattern.

The identification / recognition determination circuit 4 receives the binary image pattern of n pixels from the normalization processing circuit 2 and compares it with the binary mask pattern of each category in the binary mask pattern storage circuit 3 to, for example, The black part (or white part) of the input image pattern is included in the black part (or white part) of the binary mask pattern of the specific category, and the binary mask pattern black part (or white part) and the binary of the other category If it is not included only in the overlapping part with the black part (or white part) of the mask pattern, it will be identified and recognized as that particular category,
If the black portion (or white portion) of the input image pattern is included only in the overlapping portion, the image is rejected. Otherwise, the similarity or the like is known based on the white or black portion of the input image pattern. Or a recognition function using a recognition function that recognizes white characters or the like with high accuracy, and outputs a recognition result.

The binary mask pattern creation circuit 5 receives the image information output from the image input / output device 6 and the normalized binary image pattern of the added / updated or newly created category from the normalization processing circuit 2. Create a binary mask pattern. The binary mask pattern created by the binary mask pattern creation circuit 5 is newly registered in the binary mask pattern storage circuit 3, or added to or updated in the contents of the binary mask pattern storage circuit 3.

The image input / output device 6 inputs classification result information such as a character or graphic category number, a discrimination function value, and a binary image pattern output from the identification / recognition determination circuit 4, and displays the information. In addition, for example, image information is input and displayed using a keyboard, a mouse, or the like. further,
The image input / output device 6 sends information required for newly creating or updating a binary mask pattern to the binary mask pattern creating circuit 5.

First, the identification and recognition processing of an input image pattern will be described with reference to FIG. After the binary image pattern of the character to be identified / recognized is read out from the binary image pattern storage circuit 1 and the position and size of the character image pattern are normalized in the normalization processing circuit 2, the identification is performed. Input to the recognition determination circuit 4; here,
The character image pattern normalized by the normalization processing circuit 2 is referred to as a normalized character pattern X. Normalized character pattern X _{is, X = (x 1, x} 2, ..., x i, ... x n) can be represented by a vector having made n elements, each element x _i is the image position i (= 1, 2 ,..., N) (eg, x _i =
“0” is a white point, and x _i = “1” is a black point).

The identification / recognition determination circuit 4 stores the normalized character pattern X in the j-th mask pattern storage circuit 3.
The k-th character (k = k) of the character category (j = 1, 2,..., J)
1, 2,..., K _j ) are compared with the binary mask pattern M _jk . M _jk is _{M jk = (m 1, m} 2, ... m i, ..., m n) is represented by _jk, each m _i takes a binary, for example m _i = "0" if the white point, m _i = “1” indicates a black point.

As an example of the comparison, a case where the normalized character pattern X and the binary mask pattern _Mjk are compared will be described.
FIG. 2 shows an example of the processing flowchart. First,
Between the normalized character pattern X of the elements x _i and binary elements m _i of the mask pattern M _jk, and the number A of the _{(a) x i = "1} ", and _{(b) m i = "1} " The number B _jk and (c) x _i
= “1” and the number C _{jk such} that m _i = “1”, and (d) x _i
= “1” and the number D _{jk for} which m _i = “0” are counted (step 101).

Next, determination processing is performed in order from condition (1) to condition (3) shown below, and identification, rejection, and recognition are determined (step 102). Condition (1): If there is only one character category j that satisfies C _jk ≦ B _jk and A = C _jk , it is determined that the input character has been identified as the character category j (step 110). Condition (2): If there are two or more character categories j satisfying C _jk ≦ B _jk and A = C _jk , it is determined that the input character is rejected (step 120). Condition (3): When there is no character category that satisfies the conditions (1) and (2), a character with a strong background noise between the normalized character pattern X and the binary mask pattern _Mjk Recognition processing is performed using an identification function for recognizing outline characters or a known identification function (step 103), and recognition or rejection is determined (step 104). If the character category is recognized, the character category is set as a candidate category (step 130).

Here, as an embodiment of the recognition processing (step 103) in the case of the condition (3), a case where a new similarity S _H (X, M _jk ) according to the present invention is used for a discrimination function will be described. That is, the number of black pixels A of the normalized character pattern X
, A similarity S _H (X, M _jk ) represented by the following equation is calculated.

[0021]

(Equation 1)

As a result of the recognition, the character category j having the highest similarity is output as the character category recognized first. In this case, since the similarity S _H (X, M _jk ) does not need to extract a feature amount related to the character line structure such as the direction of the character line and the connection relationship, the feature extraction process is simplified, and furthermore, the blur is reduced. It has a feature that even if these character line structures are deformed due to crushing, noise in the character background, or the like, the recognition performance is not easily affected. In the recognition processing in the case of the condition (3), the absolute value | S _H (X, M _jk ) | of the similarity S _H (X, M _jk ) may be used as the identification function.

Next, the specific processing of the identification / recognition / rejection determination processing will be described with reference to FIG. FIG. 3 shows that n = 16
(Pixel), 2 character category (j = 1, 2), K ₁ = 1,
This shows an example of a normalized character pattern X and a binary mask pattern M _jk when K ₂ = 2. Numeral 11 denotes a binary mask pattern M ₁₁ (the first binary mask pattern of the first character category), and the number B ₁₁ = 8 when mi = "1". Reference numeral 12 denotes a binary mask pattern M ₂₁ (the first binary mask pattern of the second character category), and m _i = “1”
Is B ₂₁ = 7. 13 is a binary mask pattern M ₂₂ (first second binary mask pattern of the second character category), a number B ₂₂ = 8 serving as m _i = "1". 2
1 to 26 are six examples of the normalized character pattern X. It should be noted that the normalized character pattern 25 is a white character pattern obtained by inverting the normalized character pattern of the normalized character pattern 24 in black and white.

First, the normalized character pattern X of the normalized character pattern 21 and each of the binary mask patterns 11 to 13
The case of collating with the value mask pattern _Mjk will be described. x
_{When the} number A = 5 where _i = “1”, x _i = “1” and m _i =
The numbers that become “1” are C ₁₁ = 5, C ₂₁ = 4, C _{22, respectively.}
= 3, only the binary mask pattern M ₁₁ of the binary mask patterns 11, to satisfy the condition (1), the normalized character pattern X of the normalized character pattern 21 is identified in the first character category.

Next, in the case of the normalized character pattern X of the normalized character pattern 22, A = 6, x _i = “1” and m _i
= “1” are respectively C ₁₁ = 5, C ₂₁ = 6, C
₂₂ = 5, only the binary mask pattern M ₂₁ is, to satisfy the condition (1), the normalized character pattern X of the normalized character pattern 22 is identified in the second character category.

Next, in the case of the normalized character pattern X of the normalized character pattern 23, A = 4, x _i = “1” and m _i
= “1” means that C ₁₁ = 4, C ₂₁ = 4, C
₂₂ = 3, and the binary mask pattern M _{11 of the} binary mask pattern ₁₁ and the binary mask pattern M _{21 of the} binary mask pattern 12 satisfy the condition (2). X is rejected.

In the case of the normalized character pattern X of the normalized character pattern 24, A = 7, x _i = “1” and m _i
= “1” are respectively C ₁₁ = 5, C ₂₁ = 6, C
₂₂ = 6, and none of the binary mask patterns 11 to 13 satisfies the conditions (1) and (2), so that the identification function of the condition (3) is executed.

In this case, the similarity S _H between the normalized character pattern X of the normalized character pattern 24 and each of the binary mask patterns M _jk of the binary mask patterns 11 to 13 is:

[0029]

(Equation 2)

The next, the binary mask pattern M ₂₁ is most similarity of the binary mask patterns 12 is increased, the second character category is the recognition result of the first position.

Finally, in the case of the normalized pattern X of the normalized pattern 25, that is, the white character pattern obtained by inverting the normalized character pattern 24 in black and white, A = 9 and x _i =
The numbers of “1” and m _i = “1” are C ₁₁ =
3, C ₂₁ = 1, C ₂₂ = 2, binary mask pattern 1
Since the binary mask pattern M _jk of any 1-13 also conditions (1) and does not satisfy the (2), the identification function of the condition (3) is executed. Here, instead of the similarity S _H (X, M _jk ), the absolute value | S
_{When H} (X, _Mjk ) | is used, the normalized character pattern 2
5 normalized character pattern X and binary mask patterns 11 to 11
13 with each binary mask pattern M _jk | S _H |
Is

[0032]

(Equation 3)

The binary mask pattern M _{21 of the} binary mask pattern 12 has the highest similarity, and the same applies to the normalized character pattern 25 which is a black-and-white inverted pattern of the normalized character pattern 24. The character category is the first recognition result.

Next, another embodiment of the recognition process (step 103) for the above condition (3) will be described. In the following embodiment, as a discriminant function in the recognition process in the case of the condition (3), another novel similarity S according to the present invention represented by the following expression is used.
_c (X, M _jk ) is used.

[0035]

(Equation 4)

The term of the numerator of the new similarity S _c (X, M _jk ) is the number of C _jk and x _satisfies x _i = “1” and m _i = “1” in the input and the binary mask pattern. _i = "0" and m _i =
From the product of the number that becomes “0” (n−B _jk −D _jk ), x _i
= "0" and the value obtained by subtracting the product of the m _i = "1" and becomes the number (B _jk -C _jk) and x _i = "1" and m _i = "0" becomes the number D _jk, C _jk (n−B _jk −D _jk ) − (B _jk −C _jk ) · D _jk = n ·
This is derived from C _jk -A · B _jk . Normalized character pattern 2 shown in FIG.
4 normalized character pattern X and each binary mask pattern 11
The similarity S _c of each of the binary mask patterns M _jk to 13 is

[0037]

(Equation 5)

As in the case where the similarity S _H (X, M _jk ) is used, the binary mask pattern M _{21 of the} binary mask pattern 12 has the highest similarity, and the second character category is the first character category. The result of the recognition of the position is obtained.

A normalized character pattern 25 which is a white character pattern obtained by inverting the normalized character pattern 24 in black and white
In the case of the normalized character pattern X, the similarity S _c (X, M
_jk ), the normalized character pattern 25 and each of the binary mask patterns 11 to 13 are used when the absolute value | S _c (X, M _{j k} ) |
The similarity | S _c |

[0040]

(Equation 6)

The binary mask pattern M _{21 of the} binary mask pattern 12 has the highest similarity, and the normalized character pattern 25 is the same as the normalized character pattern 24.
, The second character category is the first recognition result.

Next, another method of the identification / rejection determination processing will be described. In this method, in the determination of the above conditions (1) to (3), the new similarity S
Processing is performed using _c (X, M _{j k} ). In the following description, conditions corresponding to the above conditions (1) to (3) are referred to as conditions (1a) to (3a), respectively.

In the identification / recognition / rejection determination process, the element x _i of the normalized character pattern X and the binary mask pattern M _jk
The elements A, B _jk , and C _jk are counted between the elements m _i (see step 101), the similarity S _c (X, M _jk ) is calculated, and stored in the binary mask pattern storage circuit 3 in advance. The processing of the following conditions (1a) to (3a) is sequentially performed using the stored identification threshold value t _jk of each binary mask pattern (see step 102). The details of the method of setting the identification threshold value t _jk of the binary mask pattern will be described later.

Condition (1a): If there is only one character category j that satisfies S _c (X, M _jk ) ≧ t _jk , it is determined that the input character has been identified as character category j (step 110). reference).

Condition (2a): If there are two or more character categories j satisfying S _c (X, M _jk ) ≧ t _jk , it is determined that the input character is rejected (see step 120).

Condition (3a): If there is no character category that satisfies condition (1a) or condition (2a), the normalized character pattern X and the binary mask pattern _Mjk
A recognition process is performed using the similarity S _c as a discriminant function (see step 103). If there is a character category j that satisfies the similarity S _c (X, M _jk ) ≧ u _jk , The character category of the binary mask pattern _Mjk is output as the first recognition result (see step 130). On the other hand, if there is no character category j that satisfies the similarity S _c (X, M _jk ) ≧ u _jk , that is, the similarity S _c (X, M _jk ) <u _jk for all the binary mask patterns M _jk If so, it is determined that the recognition is rejected (see step 120). Incidentally, the variable u _jk for each binary mask pattern M _jk, is the threshold for rejection determination that is set in advance, even smaller than the identified threshold value t _jk (u _{j k} <
t _jk ).

Here, the specific processing of the identification / rejection determination processing will be described again with reference to FIG. _First , a case where the normalized character pattern X of the normalized character pattern 22 is compared with the binary mask patterns Mjk of the binary mask patterns 11 to 13 will be described. The similarity S _c between the normalized character pattern 22 and each of the binary mask patterns 11 to 13 is

[0048]

(Equation 7)

The next, only the binary mask pattern M ₂₁ is to satisfy the condition (1a), the normalized character pattern X of the normalized character pattern 22 is identified in the second character category.
Next, in the case of the normalized character pattern 26, the similarity S of each of the binary mask patterns 11 to 13 with each of the binary mask patterns is set.
_c is

[0050]

(Equation 8)

The next, since the binary mask pattern M ₁₁ and binary binary mask pattern M ₂₁ of the mask pattern 12 of the binary mask pattern 11 satisfies the condition (2a), the normalized character pattern X of 26 is rejected . In the case of the normalized character pattern 23, each of the binary mask patterns 11 to 13
The similarity S _c with

[0052]

(Equation 9)

Since none of the binary mask patterns 11 to 13 satisfy the conditions (1a) and (2a), the similarity S _c of the condition (3a) is satisfied.
Is executed. In condition (3a), because of the high binary mask pattern M ₂₁ is most similarity of the binary mask pattern 12, the second character category is the recognition result of the first position.

Next, a process of newly creating a binary mask pattern from erroneously recognized characters or unlearned characters and adding / updating the contents of the binary mask pattern storage circuit 3 will be described.
A few embodiments will be described.

<Embodiment 1> In this embodiment, a binary mask pattern is formed by iterative learning using a similarity S represented by the following equation as a known discrimination function based on information on a black portion or a white portion of an input image pattern. This is an example of creating.

[0056]

(Equation 10)

First, information relating to a character category to be added / updated or newly created (added here) (hereinafter referred to as a j-th character category) is input to the binary mask pattern creation circuit 5 from the image input / output device 6. Is done. The binary mask pattern creation circuit 5 obtains N input image patterns (normalized character patterns) X ₁ belonging to the j-th character category to be learned and obtained from the normalization processing circuit 2 for the input image to be added. , X ₂ ,..., X _N , a new binary mask pattern M _{j (k + 1)} is created in the following steps.

(Step 1) N input image patterns X
_1, X _2, ..., the X _N, which counts the total number A _i of the input image pattern X as the x _i = "1" a total of A _T and the pixel position i of the number A of the x _i = "1" . Sorted in descending order of A _i, the column _{A '1, A' 2,} ..., A 'l, ..., A' by placing the _n, with respect to threshold a previously set (0 <a <1) hand

[0059]

[Equation 11]

L that satisfies is obtained. However, when N is small, for example, when N = 1, a = 1.

(Step 2) For an image position i corresponding to each rank of l = 1, 2,..., L, the element of the binary mask pattern M _{j (k + 1)} is set to m _i = “1”. Image positions i other than are set to m _i = “0”.

(Step 3) N input image patterns X
_1, X _2, ..., created in the image X _t and Step 2 of X _N 2
Similarity S ( _Xt , M ₎ with the value mask pattern _{Mj (k + 1).
j (k + 1)} ) to obtain an image _Xt having the maximum similarity.
The image _Xt is replaced with a new binary mask pattern M _{j ( k + 1)} .

(Step 4) Again, for the N input images X ₁ , X ₂ ,..., X _N , each image X _l and all the 2 Xs already stored in the binary mask pattern storage circuit 3
The similarity S between the value mask pattern _Mjk and the binary mask pattern _{Mj (k + 1)} created in step 3 is calculated and recognized. If all N images are recognized as the j-th character category, go to step 6, otherwise go to step 5.

(Step 5) For the M (<N) images correctly recognized in the j-th character category, the binary mask pattern M
_{j (k + 1)} is created and stored in the binary mask pattern storage circuit 3. For the remaining (N−M) images, k = k +
1. Return to step 1 as N = N−M.

(Step 6) A binary mask pattern M _{j (k + 1)} is created in the same procedure as in steps 1 and 2, and stored in the binary mask pattern storage circuit 3 to end the processing.

<Embodiment 2> This is based on the information on the black part or the white part of the input image pattern (sample pattern), and by using the similarity S _H by iterative learning, a new binary mask of each character category is obtained. It is an example of creating a pattern and an identification threshold.

First, information relating to a character category to be newly created (referred to as a j-th character category) is input to the binary mask pattern creation circuit 5 from the image input / output device 6.
The binary mask pattern creation circuit 5 applies a character category j = 1, 2,..., J to be newly created, obtained from the normalization processing circuit 2, to the input image.
In the following steps, K _j binary mask patterns M _j1 , M
_{j2, ..., M jk, ...} , to create a M _{j k j.}

(Step 1a) The number of binary mask patterns of the j-th character category to be newly created is set to K _j = 1. For N _j input image patterns (normalized character patterns) X ₁ , X ₂ ,... X _Nj belonging to the j-th character category,
x _i = “1” at the number A _T where x _i = “1” and the pixel position i
Counting the number A _i to be. Then, x _{i at} pixel position _i =
The ratio A _i / N _{j of the} number of “1” and x _i =
The binary mask pattern M _j1 is created by comparing the ratio A _T / (nN _j ) of “1” with the following equation. If A
_i / N _j ≧ A _T / if (nN _j), setting the elements of M _j1 to m _i = "1" otherwise, set the element of M _j1 to m _i = "0" In addition, the input image pattern (Normalized character pattern) X ₁ ,
A label j1 is assigned to X ₂ ,..., X _Nj as the pattern used to create the binary mask pattern M _j1 . The above processing is performed for all character categories to be newly created.

(Step 2a) For all input image patterns X _t to be newly created, the similarity S _H (X _t , M _jk ) is calculated between all the binary mask patterns M _{jk of} all character categories. _Find M _lh having the maximum similarity. Now, it is assumed that the label jk is already assigned to the input image pattern X _t. If, if l = j, X _t is determined that can be correctly recognized by the j-th character category, assign a new label jh. Otherwise, it is determined that _Xt has been erroneously recognized, and a new label je is assigned. After this process is completed, if for all input image patterns X _t, unless erroneous recognized and determines X _t is, the flow proceeds to step 4a. Otherwise, go to step 3a.

(Step 3a) h = 1 is set for the j-th character category to be newly created. If there is one or more input image patterns for all the input image patterns to which the label jk is assigned, a new binary mask pattern M _jh is created in the same procedure as in Step 1a, and h = h + 1 is set. . This determination process is performed for all K _j labels of the j-th character category. Next, If the input image pattern is present at least one assigned a label je, for all their input image patterns X _t, the same procedure as step 1a, the temporary binary mask pattern M create _je and all x _t
A similarity _{S H (X t, M je} ) was calculated with the M _je, the input image pattern X _S having the maximum similarity as a novel binary mask pattern M _jh, put the h = h + 1. Let K _j = h be the number of new binary mask patterns for the jth character category.
far. This process is repeated for all character categories to be newly created, and a new binary mask pattern is stored in the binary mask pattern storage circuit 3. Then, the process returns to step 2a.

(Step 4a) Each binary mask pattern M
_jk and between all the input image patterns X _t other than the j character category, the similarity _{S H (X t, M jk} ) was calculated to obtain the maximum similarity t _jk for each binary mask pattern M _jk , The maximum similarity is determined by the identification threshold t of the binary mask pattern _Mjk.
jk , and this is stored in the binary mask pattern storage circuit 3. When these processes are completed for all the binary mask patterns, the process ends.

<Embodiment 3> This is similar to Embodiment 2 using the similarity S _H , but uses a binary mask pattern and an identification threshold using an image pattern of a character that has been incorrectly recognized or an unlearned character. This is an example of updating a value. First, information relating to a character category to be updated (referred to as a j-th character category) is input from the image input / output device 6 to the binary mask pattern creation circuit 5. The binary mask pattern creation circuit 5 applies a binary mask to the input image in the following steps for each of the character categories j = 1, 2,..., J to be updated, obtained from the normalization processing circuit 2. Update the pattern.

(Step 1b) The number of binary mask patterns of the j-th character category before updating is set to K _j . For all the input image patterns X _t be updated, all the binary mask pattern M _jk of all character category (j = 1,2 ..., J) (k =
The similarity S _H (X _t , M _jk ) is calculated with _respect to 1, 2,..., K _j ) to find M _lh having the maximum similarity. Now, the input image pattern X _t is to belong to the j-th character category. If, if l ≠ j, it is determined that X _t has been incorrectly recognized and assigns a new label je. After this processing is completed for all the character categories to be updated, if there is no _Xt that determines that all input image patterns _Xt have been erroneously recognized,
Proceed to step 3b. Otherwise, go to step 2b.

(Step 2b) For all the input image patterns X _t to which the label je is assigned, a temporary binary mask pattern M _je is created in the same procedure as in step 1a of the second embodiment, and all the X _t and M similarity _{S H (X t, M je} ) between _je was calculated, binary mask pattern should be added to the input image pattern X _s having the maximum similarity M _{j (Kj + 1)}
Is stored in the binary mask pattern storage circuit 3. Also, K _j = K _j +1 is set, and the process returns to step 1b.

(Step 3b) Each binary mask pattern M
If there is _jk and all the input image patterns X _t other than the j character category, the similarity _{S H (X t, M jk} ) was calculated between the X _t, the maximum for each binary mask pattern M _jk When the similarity t ′ _jk is obtained and this maximum similarity is larger than the identification threshold value t _jk (t _jk = −n when newly created) stored in the binary mask pattern storage circuit 3 Is the identification threshold value t _jk = t ′ of the binary mask pattern M _jk
jk and update the new value to the binary mask pattern storage circuit 3
To memorize. When these processes are completed for all the binary mask patterns, the process ends.

Here, in the second and third embodiments, the case where the similarity S _H is used as the discrimination function has been described. However, the absolute value of the similarity S _H can be used as the discrimination function. .

In addition, as a method of creating a binary mask pattern, other methods based on known statistical inference (for example, “Statistics” written by Tetsuo Suzuki (Asakura Shoten) or “Pat
tern classification and scence analysis " O.
Duda and P.S. E. FIG. By Hart (John Wiley & Sons I
ns.)). Also,
Also in the first embodiment, when there is an image pattern that is not included in any character category, a binary mask pattern of a new character category can be created using the image pattern.

Further, in the above description, characters are mainly shown. However, general binary images such as figures can be processed in the same manner. As an example, a case where the present invention is applied to a method of recognizing a disease category based on a diagnosis result of multiple items used in medical treatment or the like will be briefly described below. For example, the results of a plurality of diagnostic items are represented on a check sheet as a binary image pattern as shown in FIG. 3 (for example, if the item is blood pressure, it is “1” if the blood pressure is high, and “0” if it is low. The pattern on this check sheet is input as a binary image pattern and stored in advance based on typical diagnostic data of the category of each disease name. The category of the disease name is recognized by collating with the binary mask pattern thus set. In this way, after the recognition target is converted into a binary image, the recognition is performed as in the above-described embodiment, whereby the recognition performance of the disease name category can be improved.

[0079]

As described above, according to the present invention,
By comparing the input image pattern with a binary mask pattern previously stored in a standard dictionary to identify, reject, or recognize the input image, the OCR user can sequentially visually inspect the erroneously recognized or rejected pattern. In the correction work, it is not necessary to re-examine the identified image, which is advantageous in that the work load is reduced.
Also, in the conventional character recognition method, one element of the recognition feature is often stored at a depth of n (n> 1) bits. However, in the present invention, since each element of the binary mask pattern can be represented by 1 bit, The storage capacity can be reduced to 1 / n. Further, by using a known image coding method together, the storage capacity of the binary mask pattern can be further reduced, and the processing efficiency is high and the dictionary capacity is compact.

In any of the first method, the second method, and the third method described in the section of the means for solving the problems, the characteristics relating to the character line structure such as the direction of the character line and the connection relationship. Since there is no need to extract the amount, the feature extraction process is simplified, and furthermore, there is an advantage that even if these character structures are deformed due to blurring, crushing, noise of the character background, etc., the recognition performance is not affected. . Even if a binary mask pattern created from an image pattern composed of a black point in the character portion and a white point in the background is used as well as a blur, crush, and noise in the character background, the character portion is configured as a white point and the background in a black point There is an advantage that high recognition performance can be obtained even for blurring, crushing, and background noise of so-called white characters.

According to the first method, an OCR user can identify, reject, or recognize an input image by comparing the input image pattern with a binary mask pattern stored in advance in a standard dictionary. In the work of sequentially revising and correcting misrecognition and reject patterns visually,
There is no need to re-examine the recognized image, and there is an advantage that the work load is reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration example of an image pattern identification / recognition processing device according to the present invention.

FIG. 2 is a flowchart of an example of identification / rejection / recognition determination processing according to the present invention.

FIG. 3 is a diagram showing an example of a pattern for explaining specific determination processing of identification, rejection, and recognition.

[Explanation of symbols]

 Reference Signs List 1 binary image pattern storage circuit 2 normalization processing circuit 3 binary mask pattern storage circuit 4 recognition determination circuit 5 binary mask pattern creation circuit 6 image input / output device

Claims (4)

(57) [Claims] 1. An input binary image pattern,
An image pattern identification / recognition method for determining a recognition category to which a binary image pattern belongs by comparing each binary mask pattern for each recognition category that has been created in advance, wherein a black portion of the input binary image pattern is counted. A first step, a second step of counting the number of black portions of the input binary image pattern included in the black portions of the binary mask pattern for each of the recognition categories, and a number counted in the second step (hereinafter, referred to as A third step of comparing Ci (where i is the number of categories) and the number of black portions for each of the recognition categories (hereinafter Bi (where i is the number of categories)) and the number counted in the first step (hereinafter referred to as Bi) A) and Ci, and there is only one recognition category i where Ci ≦ Bi in the third step and A = Ci in the fourth step. If not, the image pattern identification and recognition method characterized in that it comprises identification procedure the input binary image pattern and belongs to the recognition category i. 2. A rejection procedure for rejecting the input binary image pattern when there are a plurality of recognition categories i in which Ci ≦ Bi in the third step and A = Ci in the fourth step. The image pattern identification / recognition method according to claim 1, wherein: 3. An input binary image pattern,
An image pattern identification / recognition method for determining a recognition category to which a binary image pattern belongs by comparing with a previously created binary mask pattern for each recognition category, wherein a white portion of the input binary image pattern is counted. A first step of counting the number of white portions of the input binary image pattern included in a white portion of the binary mask pattern for each of the recognition categories, and a second step of counting the number of white portions of the binary mask pattern for each recognition category. A third step of comparing the number (hereinafter referred to as Ci (i is the number of categories)) with the number of white portions (hereinafter referred to as Bi (i is the number of categories) for each recognition category), and counting in the first step. A fourth step of comparing Ci with the calculated number (hereinafter referred to as A), and there is only one recognition category i in which Ci ≦ Bi in the third step and A = Ci in the fourth step. If not, the image pattern identification and recognition method characterized in that it comprises identification procedure the input binary image pattern and belongs to the recognition category i. 4. A rejection procedure for rejecting the input binary image pattern when there are a plurality of recognition categories i in which Ci ≦ Bi in the third step and A = Ci in the fourth step. 4. The image pattern identification / recognition method according to claim 3, wherein: