JPH06282690A - Character recognizing device - Google Patents

Abstract

PURPOSE:To provide the character recognizing device which can improve the rate of recognition by registering optimum word information by a user in a knowledge dictionary. CONSTITUTION:A knowledge processing part 16 sends the knowledge processing result based on the knowledge dictionary corresponding to the recognized result of a recognition part 12 to a control part 18 along with word registration information and the control part sends instructions to a display part 20 along with a character string so as to change the display depending on the case that words in the character string obtained as the result of knowledge processing are the words of word information existent in the knowledge dictionary from the beginning and the case that words are the words of word information registered by the user. Since the display part 20 changes the display method of words based on this display instruction, even when the erroneous post-processing result is outputted to a spot originally obtained with the correct answer words by the influence of word information newly registered, it is easily noticed and as a result. the optimum word information can be registered in the knowledge dictionary.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition device having improved reading accuracy so that a form or a document can be processed accurately.

[0002]

2. Description of the Related Art Conventionally, a knowledge dictionary has been used to improve the recognition rate of handwritten characters. The information held by the knowledge dictionary includes the information of the words to be recognized, context information,
And other information. The user of the character recognition device can also newly register these pieces of information. Less than,
As a conventional character recognition technique using word information of a knowledge dictionary, for example, a technique disclosed in a document: Proc.

First, a predetermined area of a form or a document is optically scanned and an optical signal from the paper surface is photoelectrically converted to obtain image data of the form or the document. Then, a character pattern to be recognized is cut out from the image data. A character to be recognized is recognized based on the cut-out character pattern, and one or a plurality of candidate characters are obtained as a recognition result.

Then, regarding the character for one unit of knowledge processing,
A character string is created by combining one or more candidate characters obtained for each character, and the candidate rank or similarity assigned to each candidate character of the character string is used to determine the distance between the character string and the word in the word information. Is calculated, and the character string having the maximum similarity is selected. Then, the word of the word information corresponding to this character string is displayed. However, when the similarity of the character string is less than or equal to a predetermined threshold value, a candidate character formed by combining candidate characters having the first candidate rank is displayed.

[0005]

In the conventional character recognition technique using the above-mentioned word information, (1) the case where the degree of similarity between the character string and the word of the word information existing from the beginning is maximum, 2) It is possible that the degree of similarity between the character string and the word of the word information registered by the user becomes maximum, but due to the influence of the word information newly registered in the place where the correct answer word was originally obtained. The word in the case of (2) may be erroneously output as the post-processing result.

[0006] For example, a candidate character obtained by recognizing a place where "Abe" is entered in the name column of a certain form (candidate rank is 2
Consider the case where the correct answer word is obtained by calculating the degree of similarity from the rank below. If the character with a candidate rank of 1 when recognizing a "part" is "gun" and the character with a candidate rank of 2 is "part", for example, it is a surname that is rarer than "Abe". If the user additionally registers in the knowledge dictionary the word "Agu" (which is not registered in the prepared knowledge dictionary), the name column of the above-mentioned form is read as "Angu". I will get out. Despite such cases, the word display in the cases (1) and (2) is the same. Therefore, the operator cannot discover the above phenomenon just by glancing at the display of the character string, so even if the performance of the knowledge dictionary deteriorates due to the word information registered later, it remains as it is and the character recognition device. There was a problem that the recognition rate of was lowered.

An object of the present invention is to solve the above-mentioned conventional problems and to provide a character recognition device which allows a user to register optimum word information in a knowledge dictionary and improve the recognition rate.

[0008]

In order to solve the above-mentioned problems, a character recognition device of the present invention includes a recognition unit for outputting a recognition result of a character pattern cut out from image data of a quantized form or document, and a knowledge unit. It comprises a word registration unit for registering word information in the dictionary, a post-processing unit for outputting a knowledge processing result based on the recognition result using the knowledge dictionary, and a display unit for displaying the knowledge processing result. In the character recognition device, when the character string output by the post-processing unit is the word of the word information that originally existed in the knowledge dictionary, and the word registered by the word registration unit in the display unit. The feature is that the word display method is changed with and.

[0009]

According to the present invention, the display on the display unit is (1) the case where the character string output by the post-processing unit is a word originally existing in the knowledge dictionary, and (2) the word registered by the user. However, the word in the case of (2) above may be erroneously output as the post-processing result due to the influence of the word newly registered in the place where the correct word was originally obtained. However, the operator can easily notice it. As a result, the user can register the optimum word information in the knowledge dictionary, and the above problem can be solved.

[0010]

Embodiments of the present invention will be described below with reference to the drawings. It should be noted that the drawings are only schematically shown to the extent that the present invention can be understood.
The arrangement position, dimensions, input / output signals, and connection relationship are not limited to the illustrated example.

FIG. 1 is a functional block diagram for explaining one embodiment of the present invention. Character recognition device 10 of this embodiment
Is a recognition unit 12 that cuts out character patterns from image data of a quantized form or document and outputs a recognition result of the cut-out character patterns, and word registration that registers word information desired by a user of the character recognition device. The unit 14, the post-processing unit 16 that outputs the knowledge processing result based on the recognition result using the knowledge dictionary, the display unit 20 that displays the knowledge processing result, and the operations of the post-processing unit 16 and the display unit 20 are described. The control unit 18 for controlling is provided. Further, in FIG. 1, reference numeral 22 is a photoelectric conversion unit that outputs quantized image data of a form or document, and 24 is an image memory that stores the image data from the photoelectric conversion unit 22.

FIG. 2 is a flow chart showing an outline of processing of this apparatus. When the process is started, first, in step S1, it is determined whether or not it is necessary to register the word information desired by the user.
If necessary, the word information is registered in step S2, and if not necessary, the process proceeds to the next process. The form or the document is read in step S3, the characters are recognized in step S4, the knowledge processing is performed based on the recognition result in step S5, and the knowledge processing result is confirmed and corrected in step S6. Then, if maintenance of the knowledge dictionary is required in step S7, the process returns to step S2 to change or delete the word information, and thereafter, the same processing is repeated. If maintenance of the knowledge dictionary is not required in step S7, the process ends.

FIG. 3 shows an example of a form. In FIG. 3, 31 is an example of a form in which an address is described, and 32.
Is an entry frame for specifying the character description area.

Hereinafter, this embodiment will be described in more detail with reference to FIGS. 1, 2 and 3. The photoelectric conversion unit 22 optically scans a predetermined reading range on a form or a document,
The optical signal L from the form or document is photoelectrically converted to output image data quantized into black and white binary, and the image memory 24 stores this image data.

The recognition unit 12 of the character recognition device 10 cuts out a character pattern from the image data in the image memory 24, and extracts various features relating to the character to be recognized from the cut-out character pattern. Then, the characteristics of the extracted character pattern are collated with the characteristics of the standard character pattern, and the recognition result and the candidate rank of the character are output. When one or more candidate characters are obtained as a recognition result for one character, and when there is one candidate character, candidate rank 1 is attached to the candidate character and output, and when there are multiple candidate characters , The candidate rank determined for each candidate character is attached to the candidate character and output.

The word registration unit 14 additionally registers the word information desired by the user of the character recognition device in the knowledge dictionary. Alternatively, the word information already registered by the user is changed or deleted (see steps S7 and S2 in FIG. 2). The knowledge dictionary has a word registration unit 1
Information for distinguishing the word information registered in 4 from the general word information prepared from the beginning is added. Hereinafter, this information is referred to as word registration information.

The post-processing section 16 performs knowledge processing using word information based on the recognition result from the recognition section 12. Post-processing unit 1
6 is a combination of candidate characters of each character for one unit of knowledge processing when the recognition result of the character of one unit of knowledge processing (for example, the recognition result of the description area of the prefecture name in the form 31 shown in FIG. 3) is input. The character string is collated with the word of the word information, and it is checked whether or not the word corresponding to the character string of the candidate characters exists in the word information. Then, when the character string A that matches the word of the word information is detected from the character strings formed by combining, the evaluation value J of the character string A is calculated. Where S
Be the sum of the candidate ranks given to each candidate character in the string, and N
Let be the total number of characters that make up the string,
The evaluation value J can be expressed as, for example, J = S ÷ N.

Whether or not the word and the character string A match is determined by, for example, whether or not the character codes of the characters at corresponding positions of the word and the character string A all match. Then, the character string A having the smallest evaluation value J is selected from the character strings A as the knowledge processing result when all the character strings formed for one unit of knowledge processing have been matched with the word information. A is sent to the control unit 18 together with the word registration information.

If only one character string A is detected when all the character strings of the knowledge processing unit sentence are matched with the words of the word information, the character string A is selected as the knowledge processing result, and The character string A is sent to the control unit 18 together with the word registration information.

If no character string A is detected when all the character strings for one unit of knowledge processing have been matched with the words of the word information, the candidate rank of each character for one unit of knowledge processing is A character string formed by combining the first-ranked candidate characters is selected as the knowledge processing result, and this character string A is sent to the control unit 18.

When the character string is detected by the word matching, the control unit 18 sends the selected character string A to the display unit 20 together with the display instruction according to the word registration information. In addition, the control unit 1
If the character string A is not detected by the word matching, 8 sends the character string A to the display unit 20 together with the same display instruction as when displaying the word of the word information registered in the knowledge dictionary from the beginning. If the character string A is a word of word information registered from the beginning, the instruction to display in the first color is
Is a word of the word information registered by the word registration unit, an instruction to display in a second color different from the first color is sent.

When the selected character string A is input together with the display instruction, the display section 18 displays the character string A on the display device by the method designated by the display instruction.

The present invention is not limited to the above-described embodiments, and therefore, the configuration, operation, processing contents, input / output signals and numerical conditions of each component may be arbitrarily changed. For example, in the above-mentioned embodiment, the user additionally registered the word information in the general knowledge dictionary in the word registration section, but another knowledge dictionary dedicated to the user is prepared and the necessary word information is registered therein. Good.

Further, in the above-mentioned embodiment, as the evaluation value J, a value obtained by dividing the sum S of the candidate ranks given to each candidate character of the character string by the total number N of the characters constituting the character string is used.
Scores corresponding to each candidate rank instead of the sum S of candidate ranks (for example, 100 points for candidate rank 1 and 90 points for candidate rank 2 are associated with lower scores as the candidate rank decreases). The sum of (corresponding) may be used. Alternatively, instead of the sum S of the candidate ranks, the sum of the appearance frequencies of the candidate characters in the character string (in this case, the appearance frequency is held in advance by the recognition unit) may be used.
Alternatively, instead of the sum S of the candidate ranks, the similarity between the candidate character and the character pattern corresponding to the candidate character may be obtained and the sum of the similarities of the candidate characters in the character string may be used. Alternatively, instead of the sum S of the candidate characters, the distance between the dictionary matrix of the candidate characters and the feature amount of the character pattern corresponding to the candidate character is obtained, and the sum of the distances of the candidate characters of the character string is used. Good. Alternatively, instead of the sum S of the candidate ranks, the sum of the appearance frequencies of the candidate characters in the character string and the sum of the candidate ranks may be used.

In addition to the above-described display method of the display unit, the display may be changed using, for example, one or more of different colors, different brightness, blinking and underlining.

Further, the post-processing section may perform knowledge processing using word information as described below in addition to the above. In order to check whether the word corresponding to the character string consisting of the candidate characters exists in the word information, the character string of the knowledge processing one unit sentence is compared with the word of the word information, and the similarity between these character strings and words is compared. Degree or inconsistency. As a word corresponding to the character string, for example, a word whose similarity to the character string exceeds a predetermined threshold value or a word whose dissimilarity to the character string does not exceed a predetermined threshold value is detected. Then, (1) when a character string whose similarity exceeds a predetermined threshold value or a character string whose degree of disagreement does not exceed a predetermined threshold value is detected,
The maximum similarity or the minimum dissimilarity is detected from the detected character strings, and the word of the word information corresponding to the character string having the maximum similarity or the minimum disagreement is used as the knowledge processing result,
Also, the maximum similarity or the minimum dissimilarity is output as the evaluation value of the knowledge processing. (2) Knowledge processing One character string whose similarity exceeds a predetermined threshold value or whose dissimilarity degree does not exceed a predetermined threshold value even if all the character strings for one unit of knowledge processing have been matched with the words of the word information.
If no individual can be detected, the knowledge processing result of the character string of the combination of the candidate characters having the first candidate rank is evaluated, and the predetermined lower limit value of the similarity or the predetermined upper limit value of the dissimilarity is evaluated. Output as a value. The lower limit value of the similarity and the upper limit value of the dissimilarity indicate that the word corresponding to the character string made up of the candidate characters does not exist in the word information.

Further, in the above embodiment, an example of knowledge processing using word information has been described, but the present invention may be applied to a character recognition device that performs knowledge processing using context information and other knowledge information.

[0028]

As described in detail above, according to the present invention, the display on the display unit is (1) the character string output by the post-processing unit is the word of the word information that originally existed in the knowledge dictionary. In the case of (2) above, there is a difference between the case of (2) and the word of the word information registered by the user, and the influence of the word information newly registered in the place where the correct word was originally obtained. The operator can easily notice that even if the word is output as the post-processing result by mistake. As a result, the user can register optimum word information in the knowledge dictionary. Therefore, it is possible to provide a character recognition device with improved reading accuracy so that a form or a document can be processed accurately.

[Brief description of drawings]

FIG. 1 is a functional block diagram showing a configuration of an exemplary embodiment of the present invention.

FIG. 2 is a flowchart showing an outline of processing of the apparatus of the embodiment.

FIG. 3 is a diagram showing an example of a form.

[Explanation of symbols]

 10 character recognition device 12 recognition unit 14 word registration unit 16 post-processing unit 18 control unit 20 display unit 22 photoelectric conversion unit 24 image memory

Claims (1)

[Claims] 1. A recognition unit for outputting a recognition result of a character pattern cut out from image data of a quantized form or document, a word registration unit for registering word information in a knowledge dictionary, and the knowledge dictionary. A post-processing unit for outputting a knowledge processing result based on the recognition result, and a display unit for displaying the knowledge processing result, wherein the display unit is a character output by the post-processing unit. Character recognition characterized in that the display method of the word is changed depending on whether the string is a word of word information that originally existed in the knowledge dictionary or a word registered by the word registration unit. apparatus.