JPH0338630B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (A) Field of Industrial Application The present invention relates to a character recognition system for recognizing handwritten Chinese characters, and more particularly to a method for determining the recognition ranking of candidate character type categories.

(b) Conventional technology Generally, in a character recognition system, the degree of similarity between a feature pattern extracted from an input character pattern and a standard feature pattern for each character type category registered in advance in a dictionary section is calculated, and Select n candidate character type categories. Then, in addition to outputting the candidate character type category with the highest degree of similarity as a recognition result, in order to correct misrecognition, the selected candidate character type categories are classified from the first to the highest degree of similarity. The recognition ranking up to nth place is determined in advance.

However, as mentioned above, using only similarity to determine the recognition ranking often results in false recognition, so after selecting multiple candidate character categories based on similarity, some post-processing is performed to determine the recognition ranking. Now I can think of a way to do that.

Conventionally, as post-processing, JP-A-59-
As disclosed in Japanese Patent Publication No. 32082, there are methods that perform grammatical processing, and methods that determine whether the characters before and after the recognized character are kanji, katakana, or hiragana, as in Japanese Patent Application Laid-open No. 59-27381. Something was proposed.

(c) Problems to be solved by the invention In the conventional technology, grammatical processing is performed as post-processing, so the knowledge section such as a grammatical dictionary becomes enormous, and furthermore, the processing content becomes extremely complicated. Furthermore, with the method of determining whether the preceding and succeeding characters are kanji or katakana, if the selected candidate character category is only kanji or hiragana, it cannot be expected to improve the recognition rate. .

Therefore, in order to improve the recognition rate in a short processing time without requiring a huge knowledge department, the applicant set priorities for each character type category according to the learning stage or frequency of school education. Then, I tried to use this priority to determine the recognition order of multiple candidate character type categories obtained by similarity calculation, but when I replaced the recognition order based on the priority for all candidate character type categories, the similarity Even if the degree of similarity is very high, if the priority is low, the recognition ranking will fall to the bottom, or conversely, even if the degree of similarity is very small, if there is no other character category with a high priority, this character type will be recognized. The category was placed at the top of the list, which caused the problem of erroneous recognition.

(d) Means for solving the problem The present invention sets a priority for each character type category and compares the degree of similarity as a calculation result with a plurality of threshold values. Categories are divided into classes, and the recognition order is changed based on priority only within this class to determine the final recognition order.

(E) Effect In the present invention, candidate character type categories with similar degrees of similarity are rearranged based on priority, and candidate character categories with similar degrees of similarity are not rearranged based on priority.

(F) Embodiment Figure 1 is a block diagram of a character recognition system to which the present invention is applied, and (1) reads characters written on an input manuscript and outputs the reading result as a binary character pattern. Character observation unit, (2) is a feature extraction unit that extracts feature patterns from input character patterns, (3) is a dictionary unit that stores standard feature patterns for each character type category, and (4) is the extracted feature patterns and standard features. This is a pattern matching unit that performs matching with patterns and calculates the degree of similarity between both patterns.

The character type categories in the dictionary section 3 are divided into categories according to frequency or learning stage of school education, and a priority is determined for each category set. For example, as shown in Figure 2, the glyph categories learned in the first to third years of elementary school are set to category sets 1 and 3a, the glyph categories learned in the fourth to sixth years of elementary school are set to category sets 2 and 3b, and the glyph categories learned in junior high school and above. All character categories are divided into three category sets, such as category sets 3 and 3c, and priorities 0 to 2 are assigned to category sets 1 to 3 in order.

The pattern matching unit 4 includes three calculation units 4a to 4c corresponding to category sets 1 to 3, respectively, and each calculation unit selects n candidate character type categories from each category set in descending order of similarity. is selected, and its character type code and the degree of similarity as a calculation result are stored in the candidate memory 5. On this occasion,
The calculation unit adds the priority of the corresponding category set to the character type code and similarity, and these three pieces of information are stored in the candidate memory 5 as information for each candidate character type category. In this way, candidate memory 5 stores n candidate character type categories from each category set, for a total of 3n candidate character type categories.

Further, in FIG. 1, reference numeral 6 denotes a knowledge section that stores the relationship between frequency and priority, and 7 refers to the contents of this knowledge section 6 to determine the 3n candidate character types stored in the candidate memory 5. It is a clustering control processing unit that determines the recognition ranking of the top n characters in the category and stores the character type codes in the result memory 8 in the order of recognition ranking. The type code is output as a recognition result to a character display device such as a word processor or a personal computer, and the character type is displayed. If the operator gives an instruction of incorrect recognition, the answer output control section 9 sequentially outputs the second and lower character type codes and controls the output so that the correct recognition result is displayed.

Next, the processing of the clustering control processing section 7 and the contents of the knowledge section 6 will be explained in more detail.

In this embodiment, the city block distance d is used as the degree of similarity, and it is assumed that the smaller this distance is, the greater the degree of similarity is. Threshold values D〓, D〓, D〓 (D〓<D〓<D〓) are stored. As shown in the flowchart of FIG. 3, the clustering control processing unit 7 first selects n candidate character type categories G from the 3n candidate character type categories stored in the candidate memory 5 in descending order of distance. ₁ , G ₂ , . . . , G _o are selected and stored in the working memory 7a in order. Then, the distances d ₁ , d ₂ , . . . for each of these n candidate character type categories are
By comparing d _o with multiple thresholds D〓, D〓, D〓, candidate character categories G ₁ to _{G o} are divided into four classes A, B, C, and D as shown in Figure 4. Classify. That is, the clustering control processing unit 7 stores, for each class, class pointers 10a to 10a indicating to which address in the working memory 7a candidate character type categories belonging to that class are stored.
10d, and as shown in the flowchart of FIG.
The contents TA, TB, TC, and TD of c and 10d are incremented when distances are compared. for example,
If candidate character type categories are stored in addresses 0-1 of the working memory 7a, B-class in addresses 2-4, C-class in addresses 5-6, and D-class in address 7, each The contents of the pointer are TA=2, TB=5, TC=7,
TD=8.

By the way, as shown in FIG. 4, the knowledge unit 6 stores in advance whether or not the order of priority can be changed in each class. It is specified that the order of priority can be changed only within the same class. Then, the clustering control processing section 7 refers to the contents of this knowledge section 6 and determines the recognition ranking.

Here, the character type code is M _I , the city block distance is d _I , and the priority is P _I (I=A, B, ..., H, K,
L, ..., R, S, T, ..., Z, D _I =0,1,
2) candidate character type category G _I as (M _I , d _I , P _I )
As shown in FIG. The relationship between city block distance d _I is d _S < d _K < d _T < d _A < d _L < d _U < d _B < d _M
<..., the working memory 7a of the clustering control processing section 7 selectively stores eight candidate character type categories G _S to G _M in descending order of distance d _I , as shown in FIG. be done.

Therefore, if all these eight candidate character type categories are rearranged based on the priority P _I ,
In the result memory 8, character type codes M _A to M _U are stored in the order shown in Fig. 5C, and the character type code M _S having the first similarity has a low priority of "2". The character code with the 6th place in terms of similarity and the 7th place in similarity.
Since _MS has a high priority of "0", it is ranked second.

However, the relationship between each city block distance d _I of the candidate character type category and the threshold values D〓, D〓, D〓 is, for example, d _S < D〓 < d _K , d _L < D〓 < d _U , d _M If <D〓, in this embodiment, since the clustering control processing unit 7 performs the classification, the contents of each class pointer 10a to 10d are TA=1, TB=
5, TC=8, TD=0, candidate character type category G _S is A class, candidate character type category G _K , G _T ,
G _A , G _L is B class, candidate character category G _U ,
G _B and _GM are classified into C class. Then, the clustering control processing unit 7 stores the character type code M _S of the A class as the first recognition rank in the result memory 8 as it is, and then stores the four character type codes M _K , M _T , M _A of the B class as is. , M _L are rearranged according to priority. That is, these priorities are
Since P _K =1, P _T =2, P _A =0, and P _L =1, the rankings are M _A , M _K , ML, and M _T and are stored in the result memory 8 in this order. Similarly, the C class character code is rearranged based on priority, and the ranks become M _B , M _M , M _U , and the result memory 8 contains the recognized characters as shown in Figure 5 D. Ranking is determined.

As is clear from comparing Figure 5 (d) with Figure 5 (c), in the present invention, the rankings are rearranged based on priority only among candidate character categories that have similar degrees of similarity. A candidate with an extremely small degree of similarity does not fall into a high recognition ranking, and a candidate with a high degree of similarity does not fall into a very low recognition ranking. That is, a ranking determination method is realized that is based on the calculation result of similarity and uses priority as an auxiliary.

Incidentally, in this embodiment, the dictionary section 3 is divided into category sets according to priorities, but it is not necessarily necessary to divide the dictionary section 3 if priority information is added to each character type category.

(g) Effects of the Invention According to the present invention, the recognition order is changed based on the priority only among candidate character type categories that have similar degrees of similarity, and the recognition order is changed based on the priority for those whose degrees of similarity are significantly different. Since no replacement is performed, an ideal ranking determination method based on similarity and supplementary use of priority is realized, and the recognition rate is improved.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a character recognition system to which the present invention is applied, FIG. 2 is an explanatory diagram showing the contents of a category set, FIG. 3 is a flowchart showing part of the processing contents of the clustering control processing section, and FIG. The figure is an explanatory diagram showing the contents of the knowledge section, and FIG. 5 is an explanatory diagram showing a specific example of recognition ranking determination. Explanation of main figure numbers, 3...Dictionary section, 4...Pattern matching section, 5...Candidate memory, 6...Knowledge section, 7...
Clustering control unit, 8... Result memory, 9... Answer output control unit.

Claims (1)

[Claims] 1 In a character recognition system that selects a plurality of candidate character categories by calculating the degree of similarity between a feature pattern extracted from an input character pattern and a standard feature pattern for each character category registered in advance in a dictionary section, In addition to setting a priority for each type category, the plurality of candidate character type categories are divided into classes by comparing the degree of similarity as the calculation result with a plurality of threshold values, and the priority is set only within the class. A recognition ranking determination method characterized in that the recognition rankings of candidate character type categories based on the candidate character type categories are swapped.