JPH06223219A - Character recognizing device - Google Patents

Abstract

PURPOSE:To facilitate the correcting work of a recognized result and to improve the performance of reliability. CONSTITUTION:An image input part 1 inputs charater image data. A character recognizing processing part 2 recognizes a character based upon character image data inputted by the input part 1. A post-processing part 3 executes the language processing of a character string recognized by the processing part 2. A dictionary part 4 is a language dictionary having a word dictionary and a grammar dictionary. A display part 5 displays a character string finally determined by the language processing of the post processing part 3. A correction part 6 allows a user to execute the correcting work of the characters displayed on the display part 5. An output part 7 outputs a determined character string (final output). At the time of outputting a recognized result, a dividing code in each word constituting the recognized character string together with the character string.

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD The present invention relates to a character recognition device, and more particularly, to a character recognition device applied to character recognition, natural language processing, and a human interface.

[0002]

2. Description of the Related Art As a known document which describes the prior art according to the present invention, there is an "optical character reading system" in Japanese Patent Laid-Open No. 4-17085. In this publication, correction is easily performed by selecting a candidate word, and image information of the entire field of the reading field is displayed in order to see the reduction of the time required for correction, and the first character of the candidate word of the matching result is displayed. A means for storing in-field character position information is provided, and only the candidate word in which the candidate word first character is positioned at the character position where the cursor is positioned is displayed and selected.
That is, the method of controlling the cursor for correcting the erroneously recognized characters is set to the word unit, and only the candidate word in which the candidate word first character is located at the character position where the cursor is positioned is displayed and selected.
The output of the character recognition result is to display the character string of the recognition result uniformly in solid writing.

Further, in the "form reading device" of Japanese Patent Laid-Open No. 3-148786, the knowledge processing is reflected in the output result in order to check the collation between the input document and the output result with high efficiency. To visualize. That is, the output of the character recognition result is such that the character string of the recognition result is uniformly displayed in solid, and for the character assumed to be erroneous recognition, it is displayed by color according to the certainty of recognition. is there.

The current character recognition technology is not yet sufficient. Therefore, it is indispensable for the user to correct the recognition result to correct characters. As a means for supporting the work, a method of outputting the recognition target document line by line, calculating the reliability of recognition, and displaying a character string that can be estimated to be erroneous recognition from the result by color is generally used. It was taken. However, the reliability performance is still insufficient, and in the end, the user has to recheck all the recognized characters. Further, in the current display method, since the recognized characters are displayed at equal intervals, it is very difficult for the user to read and correct the characters while looking at the display.

[0005]

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a character recognition device that facilitates correction work of a recognition result and improves reliability performance. It is a thing.

[0006]

In order to achieve the above object, the present invention provides (1)
A recognition unit that recognizes a character based on character image data, a language processing unit that performs a language process on the character string recognized by the recognition unit, a dictionary unit that has language information such as a word dictionary, and the language processing unit. The display unit that displays the character string finally determined by the unit, and the correction unit that can correct the character displayed by the display unit, and a word that forms the character string together with the recognized character string when the recognition result is output. Displaying the delimiter code in units, and (2) determining the position to which the delimiter code is added based on the notation length of the words that form the character string when the recognition result is output, (3)
At the time of outputting the recognition result, it is possible to change the type of the delimiter code based on the part-of-speech information of the words forming the character string, and (4) In any one of (1) to (3) above. , The delimiter code is deleted and only the recognized / corrected recognition character string is output as the final output.
Hereinafter, description will be given based on examples of the present invention.

FIG. 1 is a block diagram for explaining an embodiment of a character recognition apparatus according to the present invention. In the figure, 1 is an image input section, 2 is a character recognition processing section, 3 is a post-processing section (language processing). Part), 4 is a dictionary part, 5 is a display part, 6 is a correction part, and 7 is an output part. The image input unit 1 inputs character image data. The character recognition processing unit 2 recognizes a character based on the character image data input by the image input unit 1. The post-processing unit 3 performs language processing on the character string recognized by the character recognition processing unit 2. The dictionary unit 4 has a word dictionary and a grammar dictionary as language dictionaries. The display unit 5 displays the character string finally determined by the language processing by the post-processing unit 3. The correction unit 6 corrects the characters displayed on the display unit 5 by the user. Character string determined by the output unit 7 (final output)
Is output.

Next, the operation of the character recognition device according to the present invention will be described. As with the conventional character recognition device, you can enter an image,
Character recognition is performed and candidate characters are extracted. The post-processing unit 3 performs language processing using a language dictionary such as the word dictionary 4 to determine a correct character from the candidate characters,
Hand it over to the display. At that time, the notation length of each word and the part-of-speech information determined by the language processing are also passed to the display unit 5. The notation length is a numerical value indicating the number of characters in the notation. The part-of-speech information is a part-of-speech name such as a noun, a particle, a sahen noun, and a noun, in that order. FIG. 2 is a diagram showing an example of the notation length and the part-of-speech information.
For example, the word that composes the confirmed character string, which is a character string including misrecognized characters, is "disguise of unspecified Korean",
Determined by language processing.

The display unit 5 displays a word delimiter at the end of each word based on the notation length which is the output information. The recognized characters (fixed character string) and the word delimiter code for the number of characters indicated by the notation length are displayed. here,
The type of delimiter code to be displayed is changed using the part-of-speech information. When displaying a word with a specific part of speech, the delimiter code that follows the word is changed. For example, it is better to divide the delimiter into a suffix of a particle and auxiliary verb and a suffix of a part of speech other than that. FIG. 3 is a diagram illustrating an output example of the display unit. Here, the delimiter code is "_" for the suffix of the particle and auxiliary verb, and "_" for the other symbols.
However, in practice, the developer may set it freely.
Alternatively, a space code or the like may be used. The user checks the displayed character string and corrects any recognized character error. FIG. 4 is a diagram showing an output example of the display unit after the user correction work. The output section 7 deletes the word delimiter code, and finally outputs only the character string whose correction has been confirmed by the user as shown in FIG.

[0010]

As is apparent from the above description, the present invention has the following effects. (1) Effect corresponding to claim 1: Since the word delimiter code is added and displayed on a word-by-word basis, the user performing the correction work can read the recognition sentence word by word, which makes it easier to read and Failure to cut words due to misrecognized characters makes it easier to find. (2) Effect corresponding to claim 2: It is a component for realizing claim 1. By utilizing the notation length of words,
The position where the delimiter code is added can be easily calculated. (3) Effect corresponding to claim 3: Furthermore, since a plurality of delimiters are provided and the type of word can be changed according to the part of speech of the word, the user can read the recognition sentence not only in word units but also in betting unit phrases. Therefore, the effect of claim 1 is further enhanced. (4) Effect corresponding to claim 4: This is a means for outputting only the recognized character string modified by the user as the final output. Thereby, after efficiently correcting the misrecognized character according to claims 1 to 3, the user can
It is possible to utilize other systems such as databases.

[Brief description of drawings]

FIG. 1 is a configuration diagram for explaining an embodiment of a character recognition device according to the present invention.

FIG. 2 is a diagram showing an example of notation length and part-of-speech information of the character recognition device according to the present invention.

FIG. 3 is a diagram showing an output example of a display unit of the present invention.

FIG. 4 is a diagram showing a display example of the display unit of the present invention after user correction work.

FIG. 5 is a diagram showing an output example of an output unit of the present invention.

[Explanation of symbols]

1 ... Image input section, 2 ... Character recognition processing section, 3 ... Post-processing section (language processing section), 4 ... Dictionary section, 5 ... Display section, 6 ... Correction section, 7 ... Output section.

Claims (4)

[Claims] 1. A recognition unit for recognizing a character based on character image data, a language processing unit for performing a language process on a character string recognized by the recognition unit, and a dictionary unit having language information such as a word dictionary. And a display unit that displays the character string finally determined by the language processing unit, and a correction unit that can correct the character displayed by the display unit. When the recognition result is output, the character string is recognized together with the recognized character string. A character recognition device characterized by displaying a delimiter code in units of words forming a column. 2. The character recognition device according to claim 1, wherein, when the recognition result is output, the position to which the delimiter code is added is determined based on the notation length of the words forming the character string. 3. The character recognition device according to claim 1, wherein when outputting the recognition result, the type of the delimiter code can be changed based on the part-of-speech information of the words forming the character string. 4. The character recognition device according to claim 1, wherein the delimiter code is deleted and only the recognized / corrected character string is output as a final output.