JP2932667B2 - Information retrieval method and information storage device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information storage device for electronically storing a document input as image information and a method for searching for information.

2. Description of the Related Art In recent years, an information storage device called a document file device that inputs a document or a drawing as image information and electronically stores the image information has begun to spread, mainly in a department dedicated to managing the document or the drawing.

Hereinafter, an example of the above-described conventional information storage device will be described with reference to the drawings.

FIG. 5 is a flowchart showing the operation of the conventional information storage device. FIG. 5 (a) shows the operation at the time of registering a document, and FIG. 5 (b) shows the operation at the time of retrieval.

The operation of the information storage device that operates as described above will be described in more detail below.

First, when registering a document, document information is captured as image information by an image input device such as an image scanner. The captured image information is stored in a storage device such as an optical disk device. Next, the document name of the stored image information,
Auxiliary information used for searching for a classification, a creator, a keyword, and the like is input from a keyboard, and information indicating image information corresponding to the auxiliary information is added and stored in a predetermined location.

When searching for the image information stored in this manner, a search condition for limiting the auxiliary information is input from the keyboard, and the auxiliary information stored in a predetermined location is searched for the auxiliary information that matches the search condition. When the auxiliary information of the document information to be searched is specified in this way, the corresponding document can be read. (For example, Ohm “Introduction to Office Automation,” pages 111 to 113) Problems to be Solved by the Invention However, in the operation described above, it is necessary to input auxiliary information for search without fail when registering a document. Not only does it take time to register, but when registering and retrieving documents with multiple people, it is necessary to ensure consistency and consistency of keywords among each person, and it is difficult to manage the keyword system There was a problem that.

The present invention has been made in view of the above problems, and when registering a document,
It is an object of the present invention to provide an information search method and an information storage device that can be searched later without having to input auxiliary information for the search.

Means for Solving the Problems In order to solve the above problems, an information search method and an information storage device of the present invention provide a method of forming a predetermined character string from character code information in which characters are recognized from document information input as image information. When recognizing this character string from the image information of the document that contains the character string to be searched, the character string may be erroneously recognized due to imperfections associated with the recognition algorithm. It is designed to search for a character string that matches any of the other character strings and the correct character string.

Function The present invention recognizes a predetermined character string in character code information obtained by character recognition of image information of a document by the above-described method, so that auxiliary information such as a search keyword is not added to document information. However, the search can be directly performed from the document information, and the search conditions are expanded so as to compensate for the incompleteness of the character recognition. Therefore, it is possible to avoid a search omission due to erroneous recognition.

The principle of the method of compensating for imperfect recognition will be further described.

FIG. 3 is a conceptual diagram showing a case where a character can be ideally recognized. In the figure, regions a to h surrounded by solid lines
Indicates the existence range of the pattern of the virtual characters a to h, respectively.
They respectively indicate the range of patterns recognized as characters a to h. In this case, the areas a to h are:
Each is completely included in the region A to the region H,
It is clear that the letters a to h are all correctly recognized.

FIG. 4 is a conceptual diagram showing a case where recognition is not performed correctly. Similarly to the case of FIG. 3, the regions i to p surrounded by solid lines indicate the existence ranges of the patterns of the virtual characters i to p, respectively, and the regions I to P surrounded by broken lines are: They respectively indicate the range of patterns recognized as characters i to p. It should be noted that the areas X to Z indicate areas where no character can be recognized. In this case, all the characters i to p are set to 1 in the regions I to P.
It is not completely contained and it is impossible to perform full character recognition. For example, in most cases, the character i is correctly recognized as the character i. However, if the character i is written in a pattern close to the character j or the character n, the character i is erroneously recognized as the character j or the character n, respectively. Also, the letter m and the letter o
Has a pattern area that can exist, and there is no way to avoid erroneous recognition except by using means other than pattern recognition to understand the meaning from context or the like. This can occur when different character systems are mixed. For example, the kanji character 'On' and the Jirisha character λ
And the number '0' and the letter 'O' are good examples.

How such erroneous recognition occurs for which character has a tendency to fix the recognition algorithm, and if the tendency can be grasped, the defect is compensated at the time of retrieval. It is possible. For example, the letter i
Is misrecognized by the letter j. In this case, there is a problem if the recognition result is output in the form of printing, display, or the like. However, when searching by the character i, (character i + character j + character n)
If you search with, search omission is avoided. By expanding the search range, extra items are also searched. However, if the search conditions are changed to narrow down the search, there is almost no problem. Further, in practice, the search is hardly performed by one character, but by a idiom combining several characters. Therefore, the search range is not substantially greatly expanded. For example, to search for the string 'input',
The word "On" is replaced by "On + λ", and "Power" is replaced by "Power + Power"
Even if each is enlarged, the character string is' input + λ force +
Since only “input + λ power” is expanded and “input”, “input power”, “λ power”, etc., hardly exist, the actual expansion of the search range is extremely small.

As described above, by compensating for the imperfectness of character recognition at the time of retrieval, it is possible to greatly reduce the inconvenience of retrieval omission.

Embodiment Hereinafter, a method for searching for information according to an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram of an information storage device according to a first embodiment of the present invention. In FIG. 1, reference numeral 1 denotes an image input unit which captures information as an image from a handwritten or printed document. Reference numeral 2 denotes information processing means for performing various processes for controlling input and output of information. A storage unit 3 stores information handled by the information processing unit 2 as necessary. Reference numeral 4 denotes a code input unit for inputting auxiliary information of image information input from the image input unit 1, a character string for search, and the like. Reference numeral 5 denotes a character string recognizing unit, which cuts out and recognizes characters from the image information sent from the information processing unit 2,
A character code obtained by character recognition is returned to the information processing means 2. Reference numeral 6 denotes an output unit, which outputs specific information instructed according to the instruction input from the code input unit 4, information retrieved and extracted in accordance with the instruction, and the like.

The operation of the information storage device configured as described above will be described below with reference to FIGS. 1 and 2.

First, FIG. 2 shows a flowchart of the operation of the information storage device in the present embodiment, and FIG.
FIG. 2B shows a case where document information is stored, and FIG. 2B) shows a case where desired document information is searched. When storing document information, first, a handwritten or printed document is read as image information by an image input unit 1 such as an image scanner and transferred to the information processing unit 2. The information processing means 2 arranges the format of the image information, transfers the image information to the storage means 3 and stores it as a file. Further, if necessary, auxiliary information such as a keyword is input from the code input means 4, and the information processing means 2 prepares the auxiliary information into a predetermined format, and
And store it in a predetermined location. On the other hand, the information processing means 2 also sends information image information to the character recognition means 5.
The character recognizing means 5 sequentially cuts out characters from the image information, recognizes them, and converts them into character codes. The character recognition means 5 returns the character code information obtained by the recognition to the information processing means 2. The information processing means 2 arranges the character code information in a predetermined format, and then sends the information to the storage means 3,
Record in the designated place.

When searching for desired information from the document information stored as described above, first, if necessary, restricting conditions such as keywords are limited from the code input means 4 in order to limit files to be searched. input. Of course, when targeting all files, it is not necessary to input such a restriction condition. Next, a character string to be searched is input from the code input means 4 again. This character string is transferred to the information processing means 2, and the information processing means 2 expands the range of this character string based on a built-in correspondence table. The process of expanding the range of the character string will be described in more detail.

The correspondence table is a table in which characters that may be erroneously recognized when the character recognizing unit 5 recognizes each character are associated with each character. For example, as shown in FIG. 3, assuming that a range of a pattern of characters i to p and a range of regions I to P, which are ranges of a pattern recognized as characters i to p, are distributed, The correspondence table for characters is as shown in Table 1.

This correspondence table is actually composed of character codes representing each character. Means a special code assigned to the unrecognized character.

Assuming that the character string to be searched is a character string composed of three characters of “character j character m character o”, character j is character j and character n
And letters? And the character o is replaced with the union of the characters o and m. Therefore, the search is based on 'character j character m character o''character j character m character m''character n character m character o''character n character m character m''character? Letter m letter o '' letter? This is performed by the union of six types of character strings of the character m and the character m '.

As described above, the information processing unit 2 expands the range of the character string to be searched, and reads out the character code information of the specified range from the storage unit 3 while retrieving the expanded character string. Search for a character string that matches any of them.
The information processing means 2 converts the document information extracted by the search into
It is sent to output means 6. If the output means 6 is, for example, a CRT, it is displayed on the output means, and if it is a printer, the information is printed.

As described above, according to the present embodiment, when searching for a character string from character code information obtained by character recognition of document information input as image information, the table represents the tendency of misrecognition inherent to a character recognition algorithm. By referring to the correspondence table and replacing the character string with the union of this character string and all character strings that may be erroneously recognized, the search is performed with the range expanded substantially. Even if the character recognizing means 5 misrecognizes, the frequency of prying omissions can be greatly reduced.

In the embodiment described above, the search character string is replaced with the union of other character strings that are easily misrecognized on the basis of the correspondence table, and the search is performed. When using a character code that has just been used for search, if the search is made to compensate for imperfections in recognition, erroneous recognition will not be a major problem. It is intended to substantially expand the range of the character string so as to compensate for the incompleteness and to search for it. Therefore, any means for substantially expanding the range of the character string to be searched is not particularly limited.

Effect of the Invention As described above, the present invention substantially expands a search character string to compensate for incomplete character recognition when searching for a character string from a character code obtained by character recognition of a document. By performing a search, the frequency of search omissions can be significantly reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram of an information storage device according to one embodiment of the present invention, FIG. 2 is a flowchart showing the operation of the information storage device in the above embodiment, and FIG. 3 is a conceptual diagram showing an ideal character recognition case. FIG. 4 is a conceptual diagram showing the case of incomplete character recognition, and FIG. 5 is a flowchart showing the operation of a conventional information storage device. 2 ... information processing means, 3 ... storage means, 4 ... code input means, 5 ... character recognition means.

Continuation of the front page (56) References JP-A-2-17573 (JP, A) JP-A-62-11932 (JP, A) JP-A-1-279324 (JP, A) JP-A-4-70967 (JP) , A) JP-A-62-44878 (JP, A) (58) Fields investigated (Int. Cl. ⁶ , DB name) G06F 17/30

Claims (2)

(57) [Claims] 1. Character code information in which results obtained by character recognition of information input as image information are stored.
A search method for searching for a specified character string, wherein a character set including, for each character constituting the character string, a character constituting the character string and a character which may be erroneously recognized as the character And searching the character code information by regarding a character string that matches any one of character strings obtained by selecting an arbitrary character from each character set as a specified character string. A method for retrieving information for an information storage device characterized by the following. 2. An image storage device for storing document information input as image information, a recognition device for recognizing characters included in the document information, and an auxiliary information storage for storing code information output from the recognition device. Means for searching for a specified character string in the code information stored in the auxiliary information storage means, wherein the search means performs, for each character of the character string to be searched, Set a character set that includes the characters that make up the string and characters that may misrecognize the character, and match any one of the character strings obtained by selecting any one character from each character set An information storage device, wherein a character string is regarded as matching a specified character string, and the character code information is searched.