JPH0765128A - Method for generating dictionary for type character recognition - Google Patents

Abstract

PURPOSE:To efficiently generate the dictionary for type character recognition which enables the best character recognition within a limited amount of data. CONSTITUTION:The generating method for the type character recognition for a character recognition device which recognizes type characters of plural fonts by using a predetermined amount of data consists of an initial dictionary generating process (S1) which has templates for respective categories of the respective fonts to be recognized and generates the dictionary for character recognition that enables nearly complete character recognition, and an unnecessary template deleting process (S2) which recognizes evaluation data by the character recognition device by using the generated dictionary, and selects and deletes unnecessary templates on the basis of the recognition result up to the predetermined amount of data (S3).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for creating a dictionary for recognizing print characters, and more particularly, it performs machine recognition of print characters of a plurality of fonts, that is, features are extracted from input character images and have them in advance. The present invention relates to a method for creating a dictionary for recognizing printed characters for a character recognition device that performs identification by comparing with a dictionary.

[0002]

2. Description of the Related Art Generally, when recognizing print characters in a plurality of fonts, it is attempted to recognize characters in a plurality of fonts with a dictionary having a small number of templates. Here, since the data capacity of each dictionary that can be stored in the device is limited, it is necessary to select a dictionary that can obtain a character recognition rate as high as possible within this capacity.

That is, a dictionary for a pattern recognition device is
It is desirable that the size is small in view of the memory capacity and processing speed of the device. However, the recognition ability of a type character recognition dictionary generally improves as the number of templates increases. Therefore,
Conventionally, some proper basic fonts are learned, the evaluation data is recognized, a new template is created in the category of the font with poor results, and the template is added to create the contents of the dictionary.

[0004]

However, in such a conventional method of creating a dictionary for recognizing type characters, the dictionary to be finally created is considerably affected by the font that created the first dictionary. Becomes Moreover, the conventional dictionary must be created by trial and error, which is very troublesome. Therefore, the created dictionary has a certain recognition rate within a predetermined capacity, but its content may not always be the best. That is, the selection of the template is performed by trial and error, so that the template having a better recognition rate may not be included or the unnecessary template may be included.

Therefore, the present invention provides a method for creating a type character recognition dictionary which can efficiently create a type character recognition dictionary capable of performing the best character recognition within a limited data amount. With the goal.

[0006]

In the present invention, the first means for solving the above-mentioned problems is that, as shown in FIG. A method for creating a character recognition dictionary for a character recognition device, which recognizes characters, has a template for each category of each font to be recognized, and initially creates a character recognition dictionary capable of almost complete character recognition. The dictionary creation step (S1) and the character recognition device using the created dictionary to recognize the evaluation data, and based on the result, select an unnecessary template up to a predetermined data amount (S3). Unnecessary template deletion process to delete (S
2) and is composed.

In the second means of the present invention, in the initial dictionary creating step (S1) of the first means, a template is created for each category of each font to be recognized, and the evaluation data is recognized by the dictionary. That is, a template of unrecognizable characters is created and added to the dictionary to create a character recognition character that can almost completely recognize the evaluation data.

Further, the third means of the present invention is based on the first aspect.
Alternatively, the unnecessary template deleting step of the second means (S2)
Is a process of determining a deletion candidate based on the number of times selected as the first candidate for recognition when recognizing the evaluation data, a dictionary including a deletion candidate template, and recognizing and deleting the evaluation data by a dictionary not including the template. This is to include the first step (S2-1) of performing the determination process for determining the effect of the above, and deleting from the template that is less affected by the delete.

Further, the fourth means of the present invention is based on the above first aspect.
To the unnecessary template deleting step of the third means (S2)
Is a first step (S2-1) of deleting a template that is not the first candidate for recognition when the evaluation data is recognized.
Is to be included. Further, the fifth means of the present invention is that the unnecessary template deleting step (S2) of the first to fourth means is the template to be focused on for all templates after the first step (S2-1). The evaluation process is performed by a dictionary that includes the template and a dictionary that does not include the template, and the determination process for determining the influence of deletion is performed. The template with the smallest influence is deleted from the determination result, and this process is performed with a predetermined data amount. It is decided to include the second step (S2-2) which is repeated until it becomes.

In the sixth means of the present invention, in the unnecessary template deleting step (S2) of the first to fourth means, attention is paid to all templates after the first step (S2-1). A determination process for recognizing the evaluation data and determining the effect of deletion is performed by a dictionary that includes a template that does not include the template, and a template that has a small effect is determined from the determination result in order from the smallest to the determined data amount. That is, the second step (S2-2) of deleting all the exceeding templates at once is included.

The seventh means of the present invention is the above third,
In the determination processing of the fifth and sixth means, when the evaluation data is recognized in the dictionary including the determination target template and the dictionary that does not include the determination target template, the former correct answer number, reject number, and misread number are set to C1, R1, and E1, respectively. The latter one, C2, R2, E
When 2, α, β, γ are constants and the effect of deletion is AF, AF = α (C1-C2) + β (R2-R1) + γ (E2
That is, the influence is larger in the order of larger AF determined by -E1).

[0012]

According to the first means of the present invention, first, in the initial dictionary creating step, there is a template for each category of each font desired to be recognized, and character recognition dictionary capable of substantially complete character recognition is provided. To create. Then, in the unnecessary template deleting step, the character recognition device uses the created dictionary to recognize the evaluation data, and based on the result, unnecessary templates are selected and deleted up to a predetermined data amount.

As a result, the dictionary having the best template deleted from unnecessary ones within the data amount determined on the basis of the recognition result of the evaluation data is selected from the contents of the dictionary capable of performing substantially complete recognition. , Can be created quickly without using trial and error methods. Further, according to the second means of the present invention, in the initial dictionary creating step, a template is created for each category of each font to be recognized, the evaluation data is recognized by the dictionary, and a template of unrecognizable characters is created. In addition to the creation dictionary, the character recognition character that can recognize the character almost completely with respect to the evaluation data is created. Therefore, the initial dictionary can be created with certainty and, as a result, the character with high recognition rate can be created. A recognition dictionary can be created.

Further, according to the third means of the present invention, in the first step of the unnecessary template deleting step, the deletion candidate is determined based on the number of times selected as the first candidate for recognition when the evaluation data is recognized. Processing, a dictionary including a deletion candidate template, and a determination process for recognizing evaluation data by a dictionary not including the template to determine the effect of deletion, and deleting from a template having a small effect of deletion are unnecessary, and A template that is not affected by deletion can be reliably excluded, and a character recognition dictionary with a high recognition rate can be reliably created.

Further, according to the fourth means of the present invention, since the template which is not the first candidate for recognition when the evaluation data is recognized in the first step of the unnecessary template deletion step is deleted, the processing time is reduced. Can be shortened and a character recognition dictionary can be created quickly. Further, according to the fifth means of the present invention, in all of the templates in the second step after the first step of the unnecessary template deleting step,
A judgment process that recognizes the evaluation data and judges the effect of deletion by the dictionary that includes the template of interest itself and the dictionary that does not include the template is deleted from the judgment result, and the amount of data that has been set for this process is determined. Therefore, even if the dictionary capacity is still small and the dictionary capacity is still large only by the deletion in the first step, the templates are sorted in ascending order of the delete effect. By deleting them, the dictionary capacity can be reduced in an appropriate order.

According to the sixth means of the present invention, in the second step after the first step, all the templates are made to recognize the evaluation data by the dictionary containing the template of interest and the dictionary not containing it. The determination process is performed to determine the effect of deletion, and the templates with the least effect are deleted in order from the judgment result, and all templates that exceed the specified data amount are deleted at once. Then, even if the dictionary capacity is still large, the templates can be deleted at a time in the order of decreasing influence of deletion, and the dictionary capacity can be quickly reduced.

According to the seventh means of the present invention, the determination process numerically represents the influence of deletion, and the deletion is performed based on this, so an appropriate determination can be made.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT An embodiment of a method for creating a type character recognition dictionary according to the present invention will be described below with reference to the drawings. 2 to 3
Is the first method of creating a dictionary for recognizing printed characters according to the present embodiment.
FIG. This embodiment includes the first to third and fourth to sixth means.

In this embodiment. Figure 1 shows how to create this dictionary.
It consists of three steps surrounded by a dotted line. This is an initial dictionary creating step (corresponding to S1) and an unnecessary template deleting step (S2-1, S-2) including the first step and the second step.
Is equivalent to). In the initial dictionary creating step, a dictionary that exhibits recognition ability of 100% or close to the evaluation data is created. First, step (hereinafter, ST) 01
In, a template is created for each category of all fonts included in the evaluation data. The evaluation data is recognized using this dictionary (ST02), and it is judged whether the result is 100% or a sufficient result close to it (ST0).
3) If YES, end the process of the initial dictionary creation process, NO
If so, a new template is created for the font category that could not be read and added to the dictionary (ST04). Then, the process returns to ST02, and the recognition process is performed, and the process is repeated until a dictionary that exhibits the ability of 100% or close thereto is created.

This step assumes that the dictionary creating system has a sufficient memory capacity, but if the number of templates is increased, the recognition rate can eventually reach 100%. Next, the process moves to the first process of the unnecessary template deleting process. Here, an unnecessary template that has no effect even if deleted is found and deleted.

First, the dictionary created in the initial dictionary creating step is used to recognize the evaluation data, and the result is saved (ST05). From this result, the numbers selected for the first candidate, the second, the third ... For each template are totaled (ST06), and each template is ordered in the order deemed to be most necessary (ST07). Specifically, ST
From the totaling result of 06, the order is first set according to the number of first candidates.

Then, the template with the lowest rank is selected from the undetermined templates based on the rank determined in ST07, and it is set as the judgment target template (S
(T08), the dictionary including this determination target template recognizes the evaluation data, and stores the result (ST0).
9). Next, the evaluation data is recognized using a dictionary that does not include the determination target template, and the result is saved (ST10), and S
The results of T09 and ST10 are totaled to calculate the difference in recognition ability between the two dictionaries (that is, the effect of deleting the template) (ST11).

The calculation method is as follows. When the evaluation data is recognized by the dictionary including the determination target template and the dictionary not including the determination target template, the number of correct answers, the number of rejects, and the number of misreads of the former are set as C1, R1, and E1, respectively, and those of the latter are set as C2, R2, and E2. Then, the magnitude of the influence when deleted is AF = α (C1-C2) + β (R2-R1) + (E2-
E1) Note: α to γ are defined as AF calculated by coefficients. α, β, and γ are weights that indicate how important each change of correct answer, reject, and misread is compared to others. In this embodiment, the weights are not differentiated and α = β = γ = 1.

Next, it is determined whether or not the effect of deletion calculated in ST11 satisfies a predetermined criterion (S
T12). If the standard is satisfied, the template is deleted and the template is deleted (ST13). If the standard is not satisfied, the template is not deleted and it is checked whether any undetermined template remains (ST14). Return to, and if there is no remaining, move to the next ST. This completes the first step. When the processing up to this point is completed, there should be no unnecessary templates in the dictionary. That is, any of the templates remaining at this stage has a great influence on the recognition ability. If the size of the dictionary at this point does not exceed the limit when putting it on the product, etc., adopt this content as the dictionary, and if it exceeds the limit, delete from the template that currently remains .

Therefore, in the next second step, as shown in FIG. 4, the template having a relatively low degree of necessity is deleted from the currently remaining templates so as to satisfy the size restriction of the dictionary. First, it is determined whether or not the current dictionary satisfies the size restriction, the processing that satisfies the restriction is terminated, and if not, the process proceeds to the next processing (ST15).

Next, the template to be judged is determined as in ST08 (ST16). However, here ST
The order of 07 may not be followed. Furthermore, each ST
As in 09 and ST10, the dictionary evaluation data that does not include the template to be determined and the dictionary evaluation data that does not include the template are recognized, and the results are saved (ST17, ST18). Next, similarly to ST11, the results of ST17 and ST18 are totaled, the influence of deletion is calculated and stored (ST19).

Then, it is checked whether or not an undetermined template remains (ST20). If it remains, the process returns to ST16, and if not, the process moves to the next. Next, referring to the value calculated in ST19 for each template,
Select the template that has the least effect, and select (ST2
1) Delete the template selected in ST21 (ST
22), returning to ST15, these processes (ST16 to S
T22) is repeated until the size of the dictionary satisfies the condition.

Therefore, according to the present embodiment, unnecessary templates can be reliably deleted in the order of low frequency of use in character recognition, and even if deleted, the influence is low, and a character recognition dictionary is reliably created. can do. And
At this time, since the influence of the deletion can be displayed numerically and the determination can be made, an appropriate determination can be made. next,
A second embodiment of the method for creating a type character recognition dictionary according to the present invention will be described.

This embodiment corresponds to the above-mentioned fourth means. This embodiment is the ST of the first embodiment described above.
As shown in FIG. 5, ST23 is inserted between 07 and ST08. In this ST23, all the templates with the first candidate number of 0 are deleted without performing the determination process. A template that has never been selected as the first candidate will have almost no effect even if it is deleted, so it is effective when it is necessary to create a character recognition dictionary in a short processing time.

Next, a third embodiment of the method for creating a type character recognition dictionary according to the present invention will be described. The present embodiment corresponds to the above-mentioned sixth means. In the present embodiment, as shown in FIG. 6, ST21 and S of the first embodiment described above are used.
Instead of T22, ST24 and ST25 are used, respectively. In ST24, the size of the dictionary is limited,
Calculate the number of templates that must be deleted from the current dictionary size. In ST25, deletion is performed by the number obtained in ST24, in order from the template having the smallest influence when deleted for each template.

In the first embodiment, the influence of the deletion on all the templates is checked, only one of them is deleted, and when the deletion is still necessary, the influence of the deletion is again checked on all the templates. Then, the effect of deleting all the templates is checked only once, and the templates are collectively deleted in bad order. Therefore, this embodiment is effective when it is necessary to create a character recognition dictionary in a short processing time as in the second embodiment.

In the above description, the first to third embodiments have been separately described. However, in the present invention, these embodiments may be combined to create a character recognition dictionary if necessary. it can.

[0033]

As described above, according to the present invention,
An initial dictionary creation process of creating a character recognition dictionary that has a template for each category of each font to be recognized and that allows character recognition to be performed almost completely With a character recognition device using a dictionary, the evaluation data is recognized, and based on the result, it is composed of an unnecessary template deletion step of selecting and deleting unnecessary templates up to a predetermined data amount,
Templates are deleted from the initial dictionary that can almost completely recognize print characters in order of decreasing recognition rate according to the data capacity, and the dictionary is fixed, so the best character recognition is possible within the limited data amount. The effect is that it is possible to efficiently create a printable character recognition dictionary that is not trial and error.

[Brief description of drawings]

FIG. 1 is a diagram showing the principle of the present invention.

FIG. 2 is a flowchart showing a first embodiment of a method for creating a dictionary for recognizing printed characters according to the present invention.

FIG. 3 is a flowchart showing a first embodiment of a method for creating a type character recognition dictionary according to the present invention.

FIG. 4 is a flowchart showing a first embodiment of a method for creating a dictionary for recognizing printed characters according to the present invention.

FIG. 5 is a flowchart showing a second embodiment of a method for creating a type character recognition dictionary according to the present invention.

FIG. 6 is a flow chart showing a third embodiment of the method for creating a type character recognition dictionary according to the present invention.

[Explanation of symbols]

 S1 initial dictionary creating step S2 unnecessary template deleting step S2-1 first step S2-2 second step

Claims (7)

[Claims] 1. A method for creating a type character recognition dictionary for a character recognition device for recognizing type characters of a plurality of fonts with a predetermined data amount, for each category of each font to be recognized. An initial dictionary creation step (S1) of creating a character recognition dictionary that has a template and is capable of almost complete character recognition, and a character recognition device using the created dictionary to recognize evaluation data, and the result Based on the above, up to a predetermined data amount (S3), an unnecessary template deleting step (S2) of selecting and deleting unnecessary templates, and a method for creating a dictionary for typographic character recognition. 2. The initial dictionary creating step (S1) creates a template for each category of each font to be recognized, causes the dictionary to recognize evaluation data, creates a template of unrecognizable characters, and adds the template to the dictionary. The method for creating a type character recognition dictionary according to claim 1, further comprising: creating character recognition characters capable of substantially completely recognizing the evaluation data. 3. The unnecessary template deleting step (S2)
Is a process of determining a deletion candidate based on the number of times selected as the first candidate for recognition when recognizing the evaluation data, a dictionary including a deletion candidate template, and recognizing and deleting the evaluation data by a dictionary not including the template. 3. The type character recognition dictionary according to claim 1, further comprising a first step (S2-1) of performing a determination process for determining the effect of the deletion, and deleting the template from the template less affected by the deletion. Method. 4. The unnecessary template deleting step (S2) comprises:
The type character recognition dictionary according to claim 1 or 2, further comprising a first step (S2-1) of deleting a template that has not become a first candidate for recognition when the evaluation data is recognized. How to create. 5. The unnecessary template deleting step (S2)
After the first step (S2-1), a determination process for recognizing the evaluation data by recognizing the evaluation data by a dictionary including the template of interest and a dictionary not including the template for all templates is performed. The second step (S) in which the template with the smallest influence is deleted from the determination result and this processing is repeated until the predetermined data amount is reached (S
2-2) is included, Claim 1, Claim 2,
A method for creating a dictionary for recognizing printed characters according to claim 3 or 4. 6. The unnecessary template deleting step (S2)
After the first step (S2-1), a determination process for recognizing the evaluation data by recognizing the evaluation data by a dictionary including the template of interest and a dictionary not including the template for all templates is performed. From the judgment result, the template with the smallest influence is sorted in ascending order,
5. The type character recognition according to claim 1, claim 2, claim 3 or claim 4, comprising a second step (S2-2) of deleting all templates exceeding a predetermined data amount at once. How to create a dictionary. 7. The determination process, when the evaluation data is recognized by a dictionary including a determination target template and a dictionary not including the determination target template, sets the former correct answer number, reject number, and misreading number as C, respectively.
1, R1, E1 and the latter one is C2, R2, E2,
When α, β, and γ are constants and the influence of deletion is AF, AF = α (C1-C2) + β (R2-R1) + γ (E2
The method for creating a type character recognition dictionary according to claim 3, claim 5, or claim 6, wherein the AF is determined in descending order of AF determined by -E1).