JPH0696287A - Word collation pre-processing system - Google Patents

Abstract

PURPOSE:To efficiently execute a collation by excludng a character which cannot exist as a correct solution from a reference object at the time of collating a word, in a character recognizing system for designating plural candidate characters at every character of a character-string for forming a word. CONSTITUTION:A character recognizing system 1 registers a word which can appear in an appearance word registering part 2 in advance, and accumulates it in an appearance character management table 6. At the time of collating a word, with respect to a character candidate 102 obtained as a result of recognition of a character recognizing part 3, a word collation pre-processing part 4 refers to the appearance character management table 6 and deletes the character candidate which cannot appear, and designates a limited candidate character 103. A word collating part 5 executes a word collation processing with regard to this limited candidate character.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a word collation preprocessing system, and more particularly to a word collation preprocessing system for preprocessing post-recognition word collation in a character recognition system.

[0002]

2. Description of the Related Art In a conventional character recognition system, a plurality of candidate characters are designated for each character included in a word to be recognized, and when performing word matching on a plurality of character strings, all of the given characters are given. The word dictionary is searched and collated while changing the combination of candidate characters in a brute-force manner.

[0003]

According to this conventional word matching method, all candidate characters given to each character included in a word to be recognized are matched with a word in a word dictionary by a brute force combination. Since the operation is performed, the word dictionary is searched for the candidate character that should not originally exist in the word dictionary, and there is a drawback that the collation efficiency is unavoidably deteriorated.

An object of the present invention is to eliminate the above-mentioned drawbacks and provide a word matching pre-processing method capable of performing a pre-processing in which candidate characters are limited to only those characters that can appear as a character string and remarkably improving the matching efficiency. It is in.

[0005]

In the word matching preprocessing method of the present invention, a word that is expressed as a character string including a plurality of characters and that may appear as a recognition target is limited as an appearance word and registered in advance. The word matching is performed after performing a pre-processing for deleting the word that does not exist in the character group by the appearing word from the plurality of candidate characters for each character included in the character string prior to the word matching. And a word matching means.

[0006]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of the present invention.

The character recognition system 1 of the embodiment shown in FIG.
Is an appearance word registration unit 2 which constitutes an appearance word registration unit,
The appearance character management table 6 includes a character recognition unit 3, a word matching preprocessing unit 4, and a word matching unit 5 which form a word matching unit. In FIG. 1, the word dictionary 7 and the image data character string 101 are still included. , Candidate character 102, limited candidate character 103, and candidate word 104 are shown together.

Next, the operation of this embodiment will be described.

The character recognition unit 3 uses the image data character string 1
Character recognition is performed for 01 and each character is a candidate character 1
Specify 02. In this embodiment, eight candidates are designated for each character.

In the conventional character recognition system, the word collating unit 5 directly collates the word with respect to the candidate character 102 while referring to the word dictionary to determine the candidate word 104.

In the present embodiment, the words that can exist in the image data character string 7 are registered in advance in the appearance word registration unit 2, so that the characters that can appear in the image data character string 7 are stored in the appearance character management table 6. The word collation preprocessing unit refers to the appearing character management table 6 and deletes a character that does not exist, that is, a character that should not appear, from the candidate character 102, and specifies the limited candidate character 103. Thus, in this embodiment, any character is limited to 2 or 3 character candidates.

The word collation unit 5 collates the limited candidate characters in this way, and can perform collation with significantly improved collation efficiency as compared with the conventional method.

[0014]

As described above, according to the present invention, in a character recognition system in which a plurality of candidate characters are designated for each character of a character string forming a word, only characters that should appear in the character string to be recognized are recognized. By limiting the candidate characters to
This has the effect of significantly improving the word matching efficiency.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an embodiment of the present invention.

[Explanation of symbols]

 1 Character recognition system 2 Appearing word registration unit 3 Character recognition unit 4 Word matching preprocessing unit 5 Word matching unit 6 Appearing character management table 7 Word dictionary

Claims (1)

[Claims] 1. An appearance word registration means for preliminarily registering a word, which is expressed as a character string including a plurality of characters and has a possibility of appearing as a recognition target, as an appearance word, and prior to word matching. Before word matching, characterized in that it comprises a word matching means for performing word matching after performing a pre-process of deleting a candidate character that does not exist in the character group by the appearing word from a plurality of candidate characters for each character included in the character string. Processing method.