JP3071745B2 - Post-processing method of character recognition result - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a device for recognizing and outputting optically read characters, and more particularly to a character recognition result for automatically correcting and outputting an error in a recognition result when an error exists. And a post-processing method.

[0002]

2. Description of the Related Art Conventionally, as a technique in this kind of field, for example, there has been a technique disclosed in Japanese Patent Application Laid-Open No. Hei 2-267670. According to the technology disclosed in the above document, when an unrecognizable character (reject character) exists in a word, candidate characters are automatically identified from a character table based on the characters before and after the reject character, the character arrangement and position before and after the reject character. The character recognition result is corrected by retrieving and retrieving a word which is replaced with a reject character.

[0003]

However, in the above-described conventional method, it is necessary to determine whether or not recognition is possible during the recognition process. If the positions are close to each other, there is a problem that a candidate character cannot be searched.

[0004] The present invention solves the above-mentioned problems, and it is not necessary to determine whether or not recognition is possible in the recognition process, and it is possible to search for candidate characters even when a plurality of rejected characters are close to each other. It is an object of the present invention to provide a post-processing method for character recognition results.

[0005]

In order to solve the above-mentioned problems, the present invention provides a post-processing method for character recognition results which corrects and outputs character recognition results with reference to a word dictionary prepared in advance. Based on the candidate character that is the character recognition result and the distance between the candidate character and the character pattern, create a reference word in which the first candidate characters having the smallest distance are arranged,
The word dictionary is searched to extract all words having the same length as the reference word and the largest number of characters as candidate words, and any of the candidate words is a combination of the candidate characters.
If it can be composed of
Calculating a cost value which is the sum of the distance for the candidate words that can consist, selects a candidate word the cost values is the smallest
And output all the candidate words as the set of candidate characters.
The reference word is unknown if it cannot be composed from combinations
Judge whether the word is a word or not and refer to it if it is judged as unknown word
Output a word, and if it is determined that the word is not an unknown word, and
If there are a plurality of candidate words, the reference word is output, and
If it is determined that the candidate word is not an intellectual number and the candidate word is one
In such a case, the candidate word is output.

According to the present invention, since the post-processing method of the character recognition result is configured as described above, the most distant character is determined based on the candidate character as the character recognition result and the distance between the candidate character and the character pattern. A reference word in which small first candidate characters are arranged is created, a word dictionary is searched, and a word having the same length as the reference word and having the largest number of characters is extracted as a candidate word.
After creating candidate words from candidate characters, calculate the cost value that is the sum of the distances, select and output the word with the smallest cost value from the reference word or candidate word, and output all candidate words from the candidate characters When the word cannot be created, it is determined whether the word is an unknown word. If the word is unknown, a reference word is output. Otherwise, a reference word or a candidate word is output according to the number of candidate words. Therefore, a candidate character is searched for even when a plurality of rejected characters are close to each other without determining whether or not the character is unrecognizable in the recognition process.

[0007]

Embodiments of the present invention will be described below in detail with reference to the drawings. FIG. 1 is a flowchart showing a post-processing method of a character recognition result according to an embodiment of the present invention.
FIG. 3 is a block diagram illustrating a character recognition device that performs a post-processing method of a character recognition result according to an embodiment of the present invention. FIG. 3 is a diagram illustrating an example of a candidate character and a distance according to the embodiment of the present invention. 5 is a flowchart showing a candidate word creation process in the embodiment of the present invention, FIG. 5 is an explanatory diagram of the candidate word creation process in the embodiment of the present invention, and FIG. 6 is a diagram showing an example of cost value calculation in the embodiment of the present invention.

In FIG. 2, reference numeral 11 denotes a CPU for controlling the entire apparatus, 12 denotes document reading means for reading words on a document,
13 is a character recognition means for recognizing a read character and outputting a candidate character and its distance as a recognition result, and 14 a recognition for storing a candidate character as a recognition result and a distance representing a similarity between the candidate character and the character pattern. Result storage means, 15 is a reference word creation means for creating a reference word by combining candidate characters having the shortest distance, 16 is a word dictionary, 17 is a candidate word search means for searching a word dictionary and creating a candidate word, 18 Is a cost value calculating means for calculating the cost value of the candidate word, 19 is an unknown word determining means, 20 is an output word determining means, and 21 is a result display / output means for displaying the determined output word.

Hereinafter, the processing operation of the post-processing method for character recognition results according to the embodiment of the present invention will be described with reference to FIGS. (1) Reference Word Creation (Step 1) First, before creating a reference word, a document is read by the document reading means 12, and each character of the read word is recognized by the character recognition means 13 to generate a candidate character. Together with the distance representing the similarity between the candidate character and the character pattern,
The candidate character and the distance are stored in the recognition result storage unit 14.

If the character pattern is "world" and the candidate words and distances for each character are calculated as shown in FIG. 3, the reference word creating means 15 determines whether each character stored in the recognition result storage means 14 By combining candidate characters having the minimum distance, “w
o ′ l 2 ”is created. (2) Candidate Word Creation (Step 2) Next, the candidate word search means 17 creates a “candidate word” closest to the reference word using the word dictionary 16.

First, in step 31 of FIG. 4, the word dictionary 16 is searched to determine whether there is a word having the same length as the reference word. If there is a word having the same length, the number of matching characters between the word obtained as a result of the search and the reference word is calculated in step 32. Next, in step 33, the number of matching characters is compared with the number of matching characters of the candidate word up to that time. If the number of matching characters is increasing, the candidate words up to that point are deleted in step 34, and the word of the search result is newly registered as a candidate word. If the number of matching characters is the same, the word of the search result is added to the candidate word in step 35.

In the case of the example shown in FIG. 5, the reference word "w
o ' l 2 "and the word" world "
d , "would" and "foos" are registered as candidate words. Next, an attempt is made to create each candidate word by combining candidate characters for each character pattern. (3) Cost Value Calculation (Step 3) When a candidate word can be created from the candidate characters, the cost value calculating means 18 calculates the sum of the distances of the candidate characters used at that time as a cost value for the candidate word. I do.

(4) Output Word Determination (Step 4) The output word determination means 20 selects a candidate word having the smallest cost value as an output word. In the case of the example shown in FIG.
The cost value of the candidate word “world” is 375. The cost value of the candidate word “would” is 420. As a result, the candidate word “wor” with the smallest cost value
ld ”is selected. Then, the selected word is output from the result display / output unit 21 as a result of the post-processing of the character recognition.

Next, a description will be given of a process performed when all candidate words cannot be created from candidate characters at the time of creating candidate words. For example, if the original character pattern is "UN", as a result of character recognition, "U", "V", and "J" are created as candidate characters for "U" in ascending order of distance, and similarly as candidate characters for "N". If "N" and "H" are created in
The reference word is "UN". At this time, if the candidate words obtained from the word dictionary are “AN”, “IN”, and “ON” (assuming that “UN” is not described in the dictionary), “A”, “I” Since "" and "O" are not in the candidate characters, all candidate words cannot be created from the candidate characters. In such a case, the following condition is checked by the unknown word determining means 19 to determine whether the original character is an unknown word. 1. The number of matching characters is less than half the word length. 2. There are five or more characters that did not match. 3. There are five or more candidate words.

When any one of these conditions is true, it is determined that the word to be processed is an unknown word, and the reference word is output as a processing result. In other cases, it is determined that the word is not an unknown word. If the number of candidate words is one, the candidate word is output. Otherwise, the reference word is output. Although the embodiment of the present invention has been described in the case of English words, the present invention can be applied to other languages by changing the contents of the dictionary.

The present invention is not limited to the above-described embodiment, but various modifications are possible based on the gist of the present invention, and they are not excluded from the scope of the present invention.

[0017]

As described above in detail, according to the present invention, the first distance having the shortest distance is determined based on the result of character recognition.
Create a reference word in which candidate characters are arranged, search the word dictionary, extract the word with the same length as the reference word and the number of matching characters as the candidate word, create a candidate word from the candidate characters, and add the total distance Is calculated, and the word with the smallest cost value is selected and output.When all candidate words cannot be created from candidate characters, it is determined whether or not the word is an unknown word. Since words are output, and if not, reference words or candidate words are output according to the number of candidate words, so that it is not necessary to determine whether or not recognition is possible in the recognition process, and multiple rejects Even when characters are close to each other, it is possible to search for candidate characters.

[Brief description of the drawings]

FIG. 1 is a flowchart illustrating a post-processing method of a character recognition result according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating a character recognition apparatus that performs a post-processing method of a character recognition result according to an embodiment of the present invention.

FIG. 3 is a diagram illustrating an example of a candidate character and a distance according to the embodiment of the present invention.

FIG. 4 is a flowchart illustrating a candidate word creation process according to the embodiment of the present invention.

FIG. 5 is an explanatory diagram of a candidate word creation process in the embodiment of the present invention.

FIG. 6 is a diagram illustrating an example of cost value calculation according to the embodiment of the present invention.

[Explanation of symbols]

 1 Reference word creation 2 Candidate word creation 3 Cost value calculation 4 Output word determination

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-2-118785 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G06K 9/72

Claims (1)

(57) [Claims] 1. A post-processing method of a character recognition result in which a character recognition result is corrected and output with reference to a word dictionary prepared in advance, comprising: (a) a candidate character which is a character recognition result; And (b) searching the word dictionary to find a word having the same length as the reference word and having the largest number of characters. all as a candidate word
Extraction Te, (c) union of one of the candidate words the candidate character
If it can be composed of
Calculates the cost value that is the sum of the distances for the candidate words that can be composed from and selects the candidate word with the smallest cost value
And to output, combination of (d) all of the candidate word is the candidate character
If the reference word is unknown,
And if it is determined to be an unknown word, the reference word is
Output, and if it is determined that the word is not an unknown word,
If there are multiple words, output the reference word,
If it is determined that there is no candidate word and there is only one candidate word,
Is a post-processing method for character recognition results, wherein said candidate word is output.