JPH0934888A - Character recognition method and character recognition device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To recognize a character string coincident with input character information as a character string corresponding to the input character information by comparing the input character information with a similar character string similar to the input character information. SOLUTION: A controller 18 inputs input character string data and identification information from an image reader 13 and sets them in the interior. Then the controller 18 retrieves area information database 16 with a 1st comparison retrieval section 18a. When it is decided that data are obtained from the area information database 16, monetary agency information coincident with the character string data obtained by the area information database 16 is retrieved from a monetary agency database 17 with a 2nd comparison retrieval section 18b, read by a data acquisition section 18c and fed to an output section 15. When a name of the monetary agency corresponding to the input character string is found out as a result of retrieval, a Kana (square form of Japanese syllabary) expression character string and a code corresponding to the name of monetary function corresponding to the content in an acquisition section 18c are read out of the monetary agency database 17 and fed to an output section 15.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition method and a character recognition device, and more particularly to a character recognition method and a character recognition device for recognizing handwritten characters as a character string.

[0002]

2. Description of the Related Art FIG. 5 shows a block diagram of a conventional example. In the conventional character recognition device 20 for searching financial institution information, when the financial institution name written in the form 21 is read by the image reading device 22, the financial institution database directly by the financial institution name read by the image reading device 22 by the control device 23. 24 is searched, and when there is a financial institution name that matches the financial institution name read in the financial institution database 24, information about the financial institution name is read from the financial institution database 24 and displayed on the display 25.

In the conventional character recognition device 20, if the image reading device 22 erroneously recognizes the characters written in the form 21, the matching financial institution name cannot be found in the financial institution database 24 and the financial institution name is not found. Was not recognized. For example, if the intention is to write "Mt. Fuji" on the form, but the image reading device 22 recognizes "Mt. Hatakeshi", and there is no financial institution name "Mt. Hatakeshi" in the financial institution database 24, It was an error.

[0004]

However, in the conventional character recognition device, the character string input is directly compared with the preset character string to be recognized, and the matching character string is searched for to find the character string. Since the input character string is recognized, there is a problem in that if the input character string is erroneously recognized, the character string to be recognized cannot be obtained.

The present invention has been made in view of the above points, and an object thereof is to provide a character recognizing method and a character recognizing device capable of recognizing a character string recognized differently from input character information. To do.

[0006]

FIG. 1 shows the principle of the present invention. The storage unit 1 stores similar character information similar to the recognized character string. The comparison step 2 compares the input character information with the similar character string stored in the storage means 1.

In the character string recognition step 3, when the similar character string identical to the input character information is detected in the comparison step 2, the similar character string detected is recognized as the character string corresponding to the input character information.
According to the first aspect of the present invention, when the input character information is input, the input character information is compared with the similar character string similar to the input character information, so that the character string matching the input character information is input. Can be recognized as a character string corresponding to.

According to a second aspect of the present invention, the similar character string is composed of the plurality of character strings, and the similar character information is composed of a plurality of character strings, so that the character string corresponding to the input character information can be selected from the plurality of character strings. Therefore, correct character information can be obtained. According to a third aspect of the present invention, the similar character string is composed of identification information for identifying the same character string as different character strings, so that the character string can be recognized by the input character information and the identification information. Even if is incorrect, it is possible to recognize the character string by combining the input character information and the identification information.

According to a fourth aspect of the present invention, in a character string recognition device for recognizing input character information, a storage means for storing a similar character string similar to the input character information and the similar means for storing the input character information in the storage means. A comparing means for comparing with the character information; and a character string recognizing means for recognizing the similar character string as a character string corresponding to the input character information when the comparing means detects similar character information matching the input character information. And storing similar character information similar to the character information to be recognized in the storage means, comparing the input character information with the similar character string stored in the storage means, and inputting by the character string recognition means. By recognizing the character string that matches the character information as a character string, you can select the character string corresponding to the input character information from the similar character strings. It is possible to recognize the correct string.

[0010]

FIG. 2 is a block diagram showing a first embodiment of the present invention. In this embodiment, a device for encoding a financial institution name written by hand on a form and reading data on the written financial institution will be described.

The character recognition device 11 of the present embodiment recognizes the characters read by the image reading device 13 and the image reading device 13 for reading the description items such as the financial institution name of the form 12, and the financial institution name, its kana notation, The control unit 14 converts the information such as a code and the output unit 15 that outputs the converted information recognized by the control unit 14.

The form 12 is composed of a financial institution name portion for describing the deposit / withdrawal destination financial institution name and an input information portion for describing information such as the deposit / withdrawal amount.
The information described in 3 is read. The image reading device 13 irradiates the form 12 with light and detects the reflected light from the form 12 with an image sensor or the like to form the form 12.
The characters written on the form 12 are read based on the difference in the light reflectance between the background and the written characters.

The characters read by the image reading device 13 are supplied to the control unit 14. The control unit 14 stores a character string similar to the name of a financial institution and a regional information database 16 in which discrimination information for distinguishing the business types of financial institutions such as banks, credit unions, and credit unions is stored. Financial institution database 1 in which code data is stored
7. A control device 18 for obtaining information about the financial institution described in the form 12 by using the regional information database 16 and the financial institution database 17.

The area information database 16 is a recognition character string storing a character string of a financial institution name to be recognized.
a, a business type discrimination information storage area 16b that stores the business types of banks, credit unions, credit unions, etc. in association with each stored financial institution name, and is close to the name of the financial institution to be recognized, and is easily mistaken It has a similar character string storage area 16c for storing a similar character string.

The name of the financial institution registered in the financial institution database 17 is stored in the recognized character string storage area 16a. The business category of the financial institution is stored in the business category determination information storage area 16b. The similar character string storage area 16c stores the same number of characters as the name of the financial institution, the same character as the name, and a character similar to the name.

The area information database 16 stores business type discrimination information and a name to be recognized for each similar character string. When the character string to be recognized is “Fujiumi”, similar character strings are, for example, “? Shi ??” and “? Sat?
□ ”can be set. At this time, “?” Indicates information that any character is recognized as the same character, and “□” indicates a blank.

Further, "silver" indicating a bank, "credit" indicating a credit, "group" indicating a credit group, and the like can be set as the business type discrimination information. In this way, the regional information database 16 is a database of information unique to other financial institutions, focusing on the name of the financial institution. For example, length 3 in the same financial industry
The financial institution name with the second character "samurai" is defined as "Fujitsu" among all the financial institution names. Further, if the number of characters is not specified and it is possible to make a judgment only by the character position, it is possible not to define the number of characters.

For a branch name, a database is created based on the same idea, but an accurate name can be obtained by preferentially defining a name focusing on the area.
The financial institution database 17 includes a financial institution name storage unit 17a that stores the name of the financial institution, a Kana name storage unit 17b that stores the name of the financial institution in Kana expression, and a code storage unit 17c that stores a code for identifying the financial institution. Etc.

The financial institution name storage unit 17a stores the name to be written on the form 12. The kana name storage unit 17b stores kana expression characters of the financial institution name. The code storage unit 17c stores a code preset for each financial institution. The financial institution database 17 stores, for each financial institution name, a kana expression character of the financial institution name and a code for identifying the financial institution.

The control device 18 is an image reading device 13.
Input character information and local information database supplied from
6 is compared and the preset comparison character string which is considered to correspond to the character string of the input character information is searched for by the first comparison / search unit 18a and the first comparison / search unit 18a. The second comparison / search unit 18b that obtains the information of the financial institution having the same character string by comparing the character string with the contents of the financial institution database 17 and the information of the financial institution name described in the form 12 to the financial institution database 17 The data acquisition unit 18c is configured to acquire more.

The control device 18 searches the area information database 16 and the financial institution database 17 according to the operation described later to perform character recognition and data acquisition. Output part 1
A control unit 1 includes a display, a printer, and the like.
The information of the financial institution described in the form 12 acquired in 8 is output.

FIG. 3 shows an operation flowchart of the control device 18. When the control device 18 acquires the information about the financial institution handwritten on the form 12, the control device 18 converts the handwritten information of the financial institution name obtained by reading the form 12 by the image reading device 13 into character string data and The financial institution is recognized by the identification information that identifies the business type.

In the control device 18, the input character string data and identification information are input from the image reading device 13 and set inside (step S1 _-1 ). Next, the control device 18 uses the area information database 16 as a first comparison / search unit 18a.
Search according to the procedure described later (step S
1 _-2).

Next, the control unit 18 makes a step S1._-2And day
Data is obtained (step S1)_-3). S
Step S1_-3Data from the regional information database 16
Local information database 1 if it is determined that the information has been obtained
The financial institution information that matches the character string data obtained in 6
Financial institution database 1 by 2 comparison / search units 18b
7 is retrieved and read to the data acquisition unit 18c, and the output unit 1
5 (step S1)_-Four, S1 _-Five).

Further, step S1_-3Local information database
Unable to find data corresponding to source 16
If the second comparison / search unit 18b
Database 17 that matches the input string
Search for a name (step S1) _-6).

As a result of the search of the financial institution database 17 by the second comparison / search unit 18b, if the financial institution name corresponding to the input character string can be found, the financial institution database 17 corresponds to the financial institution name corresponding to the acquisition unit 18c. The Kana representation character string and code to be read are read and supplied to the output unit 15 (steps S1 _-7 , S1 _-5 ). If the corresponding financial institution name cannot be found in the financial institution database 17, data notifying an error is output to the output unit 1.
5 (steps _S1-7 , _S1-8 ).

[0027] will be described in conjunction with FIG. 4 search operation in step S1 _-2 local information database 16. FIG. 4 illustrates a search operation flowchart of the area information database 16 by the control device 18.

[0028] When performing the search area information database 16, the controller 18 sets the input character string to the first comparison and the search unit 18a (step S2 _-1). Next, the control device 18 searches the first similar character string in the area information database by the first comparison / search unit 18a and sets it in the first comparison / search unit 18a (steps S2 _-2 , S2 _-3 , S).
2 _-4 ).

Next, the control device 18 uses the first comparison / retrieval unit 1
The input character string set by 8a is compared with the search similar character string to judge whether they match or not (steps _S2-5 , S).
2 _-6 ). Here, if the input character string and the search similar character string match, the business type information given to the next input character string is compared with the business type information given to the search similar character string, and the match or non-match _Is determined (steps S2 _-7 , S2 _-8 ).

Here, if the business type information also matches, it is determined that the character string is to be recognized, the character string to be recognized is read from the regional information database 16, and the second comparison / search is performed. It is set in the section 18b (step _S2-9 ). If the input character string does not match the search setting character string in step _S2-6 , the setting character string stored next in the area information database 16 is read (step _S2-10 ), and step _S2-3. the ～S2 _-6 repeated until strings match.

At this time, if there is no corresponding character string in the area information database 16 as a result of sequentially searching the set character string to the end, error data indicating that there is no corresponding character string in the area information database 16 is controlled ( Step S2
_-11 ). Mt. Fuji is listed as the name of the financial institution, and by marking "shin" indicating credit as the business type, when the form 12 is read by the image reading device 13, the name of the financial institution is recognized as "Todo?" The operation when the business type is recognized as credit is described.

The input character string “Todo?” And the similar character string in the regional information database 16 are sequentially compared, and the similar character string “? Sat? □” for identifying “Fujiumi Bank” in the regional information database 16 is obtained. When compared with the similar character string “? Sat? □”, the first and third characters are “?” And the fourth character is a blank “□”. You can judge that they match. When the character strings match, the financial institution business type is checked. Here, the financial institution business type of the input character string is "cred", whereas the financial institution business type of the search similar character string is "silver", Since they do not match, it is determined that the names are different from the financial institution name on the form 12.

By setting a blank "□", if the character position is blank, it can be determined that the lengths match, and if the character position is blank, it can be determined that the lengths do not match, and the number of characters can be easily determined. Next, “Mt. Fuji Shinkin” in the regional information database 16
It is not recognized here because the second character is different between "Sat" and "shi" when compared with the similar character string "? Shi? □" for identifying If "? Sat ??" is stored separately from "? Shi ??", the input character string "Todo ??" can be detected as a matching character string. Since they match, the input character string can be recognized as "Mt. Fuji".

In this way, the place where "Mt. Fuji" is entered on the form 12 is recognized by the image reading device 13 as "earth" which is easy to be mistakenly recognized as "shi", and the third character "mountain" is indefinite. Even if it is said, it can be recognized as "Mt. Fuji". Therefore, the search of the financial institution database 17 can be performed easily and surely.

Further, by checking the financial institution business type, reliability can be improved. This is, for example, when the form written as "Mount Fuji" is recognized as "Fuji?"
We cannot distinguish between "Fujitsu" and "Mt. Fuji". By using the item of the financial institution business category in the regional information database 16, the definition of the name of the credit union whose length is 3 characters and whose second character is "shi" is "Mt. Fuji" Character recognition can be reliably performed.

As described above, according to this embodiment, even in the character recognition mechanism having a poor recognition rate, it is possible to improve the character recognition rate by investigating the tendency of the recognized character and making a database. In this example,
Although the financial institution name / branch name has been described, the use range is not limited to this, and the range of use can be expanded.

[0037]

As described above, according to the first aspect of the present invention, when the input character information is input, the input character information is compared with the similar character string similar to the input character information to obtain the input character information. It has the feature that a character string that matches is recognized as a character string corresponding to input character information.

According to the second aspect, since the similar character information is composed of a plurality of character strings, the character string corresponding to the input character string information can be selected from the plurality of character strings, so that correct character information can be obtained. It has features such as According to claim 3, the character rate can be recognized by the input character information and the identification information, and even if the input character information is incorrect, the character string can be recognized by the combination of the input character information and the identification information. It has features such as possible.

According to the fourth aspect, similar character information similar to the character information to be recognized is stored in the storage means, and the input character information is compared by the comparison means with the similar character string stored in the storage means, By recognizing the character string that matches the input character information as a character string by the character string recognition means, you can select the character string connected to the input character information from similar character strings, so even if the input character information is incorrect, the correct character string It has features such as being able to recognize.

[Brief description of drawings]

FIG. 1 is a principle diagram of the present invention.

FIG. 2 is a block diagram of an embodiment of the present invention.

FIG. 3 is an operation flowchart of one embodiment of the present invention.

FIG. 4 is an operation flowchart when searching an area information database according to an embodiment of the present invention.

FIG. 5 is a block diagram of a conventional example.

[Explanation of symbols]

 1 Storage Means 2 Comparison Means 3 Character Recognition Process 11 Character Recognition Device 12 Form 13 Image Reading Device 14 Character Recognition Unit 15 Output Unit 16 Regional Information Database 17 Financial Institution Database

Claims (4)

[Claims] 1. A character recognition method for recognizing input character information, the input character information when the input character information is input, and a similar character string similar to the input character information stored in a storage unit in advance. And a character string recognition that recognizes the detected similar character string as a recognition character string corresponding to the input character information when the similar character string that matches the input character information is detected in the comparison step. A character recognition method comprising the steps of: 2. The character recognition method according to claim 1, wherein the similar character string is composed of a plurality of character strings similar to each other. 3. The character recognition method according to claim 1, wherein the similar character string is provided with identification information for recognizing the same character string as different character strings. 4. A character string recognition device for recognizing input character information, a storage unit storing a similar character string similar to the input character information, the input character information and the similar character string stored in the storage unit. And a character string recognition unit that recognizes the similar character string as a character string corresponding to the input character string information when a similar character string that matches the input character information is detected by the comparing unit. A character string recognizing device having: