JP2000057265A - Character reader - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a character reader which can shorten the process time and improve the recognition rate. SOLUTION: When image data of a 1st character frame recognized by a character recognition part 15 are, for example, 'TAISHO', a select signal SEL for selecting a character set 16a for recognition of a 2nd character frame is outputted from a character set selection part 17 to a character dictionary 16A. Then the image data of the 2nd character frame are segmented by a character segmentation part 16 and supplied to the character recognition part 15, which performs character recognition according to the character set 16a for recognizing only numbers '0' and '1'. When the 1st character frame is 'SHOWA', the 2nd character frame has its characters recognized according to a character set 16c for recognizing numbers '0' to '6'. Thus, the recognition is performed by limiting character candidates according to the recognition result of a previous character frame, so the processing time is shortened and the recognition rate is improved.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character reading apparatus for reading characters or symbols written on a form in which a format is determined in advance.

[0002]

2. Description of the Related Art FIG. 2 is a block diagram of a conventional character reading system. This character reading system includes an optical character reading device (hereinafter, referred to as “OCR”) 10 and a host computer 30. The OCR 10 includes an image scanner 11 that optically reads image information of the form 1 in a predetermined format, an image memory 12 that stores image information read by the image scanner 11,
And a format registration unit 13 in which entry positions and character types of characters and the like in the form 1 are registered in advance. Further, the OCR 10 is a character extracting unit 1 for extracting image data such as characters and marks to be read from the image memory 12 based on the information of the format registering unit 13.
4. A character recognizing unit 15 for recognizing characters and marks from the cut-out image data and outputting a code, and a character dictionary 16 in which characteristics of character patterns required for character recognition are registered. The character code output from the character recognition unit 15 is provided to the host computer 30.

FIG. 3 is a diagram showing a part of a form 1 to be read. The form 1 has an entry field 2 for entering a date of birth. In the entry column 2, select the year from the Meiji, Taisho, Showa, and Heisei periods and mark it with a mark. 2a, and fill in the tens and first places of the year with numbers.
b, 2c, character boxes 2d, 2 for entering the tenth and first places of the month
e and character frames 2f and 2g for entering the tens and first places of the day are provided.

Next, the operation will be described. As shown in FIG. 3, a mark and a number are entered in a predetermined position of the entry column 2 of the form 1 and input to the image scanner 11 of the OCR 10 in FIG. 2. The image information of the form 1 is read and stored in the image memory 12. Is stored. When the image information of the form 1 is stored in the image memory 12, the character extracting unit 1
4 operates, the image data of the mark frame 2 a is cut out from the image memory 12 based on the information of the format registration unit 13, and given to the character recognition unit 15. The character recognition unit 15 detects the position of the mark based on the information of the format registration unit 13, recognizes that the year is "Taisho", and outputs a code "2" corresponding to "Taisho", for example. Similarly, the image data of the character frames 2b to 2g is sequentially cut out from the character cutout unit 14 and is provided to the character recognition unit 15. In the character recognition unit 15, the format registration unit 1
Based on the information of No. 3, character identification is performed on the assumption that a number is described in each of the character frames 2b to 2g. Then, from the character recognition unit 15, for example, the code “12” corresponding to the date
1024 "are sequentially output.

[0005] The data "October 24, 1912" output from the character recognition unit 15 of the OCR 10 is given to a host computer 30, and the host computer 30 checks the validity of the data. Then, in the case of impossible data, for example, "Taisho 50 ...", the data is processed as a reading error or an entry error.

[0006]

However, in the conventional character reading system, when the character recognizing unit 15 recognizes a character or the like, information such as the character type described in the character frames 2b to 2g is registered in a format. It was only obtained and obtained from the part 13 alone. The host computer 30 checks the validity of the read year and year data. Therefore, the following (1)
There were problems such as (3). (1) For example, even if the year is recognized as “Taisho” from the mark frame 2a of the form 1, the numbers in the character frame 2b for entering the tens digit of the year are changed from “0” to “9”. It was necessary to recognize from ten character candidates, and it took a long time for the recognition process. (2) Similarly, when the year is “Taisho”, unnecessary characters of “2” to “9” are displayed although the number in the character frame 2b is limited to “0” or “1”. Since the candidate exists, erroneous recognition may occur and the recognition rate may decrease. (3) Even if there is an error in validity and it is not necessary to recognize characters after that, all characters to be read in the form 1 are recognized by the OCR 10 and converted into codes, and the host computer Therefore, it is difficult to reduce the time required for the recognition processing. The present invention provides the above (1) to
It is an object of the present invention to provide a character reading device that solves the problem (3) and can shorten the processing time and improve the recognition rate.

[0007]

Means for Solving the Problems In order to solve the above problems, a first aspect of the present invention is to provide a first entry box for entering characters or symbols, and a first entry box for entering characters or symbols. A character reading device for reading a form having a second entry frame for entering a character in a character set corresponding to the entered character or symbol and recognizing the character or symbol entered in the form; A character dictionary in which characteristic data for recognizing a character or a symbol to be read for each set is registered, storage means for storing image information of the form, and the image information stored in the storage means. And a cutout means for cutting out image data of the first and second entry frames. Further, the character reading device refers to a specific character set in the character dictionary selected based on the selection signal, and recognizes a character or a symbol corresponding to the image data cut out by the cutout means. Means, the first character recognized by the character recognition means.
Selecting means for outputting the selection signal for recognizing the image data of the second entry frame based on the characters or symbols in the entry frame.

According to the first aspect of the present invention, since the character reading device is configured as described above, the following operation is performed. First, from the image information stored in the storage unit, image data corresponding to the first entry frame of the form is cut out by the cutout unit and provided to the character recognition unit. The character recognition means refers to the characteristic data of the character set in the character dictionary selected by the selection signal, and recognizes a character or a symbol corresponding to the image data of the first entry frame.
The recognition result of the character recognition means is given to the selection means, and the selection means outputs a selection signal for recognizing image data of the second entry frame. Next, image data corresponding to the second entry frame of the form is cut out from the image information stored in the storage means and provided to the character recognition means. The character recognizing means refers to the characteristic data of the character set in the character dictionary selected by the selection signal for recognizing the image data of the second entry frame, and a character or symbol corresponding to the image data of the second entry frame. Is recognized.

According to a second aspect of the present invention, there is provided a first entry box for entering characters or symbols, and a word in a word set corresponding to the characters or symbols entered in the first entry box. In a character reading device that reads a form having a second entry frame and recognizes a character, symbol, or word entered in the form, characteristic data for identifying a character or a symbol to be read is registered. Character dictionary,
A word dictionary in which a plurality of words to be read are registered for each word set, storage means for storing image information of the form, and the first and second words based on the image information stored in the storage means. And a cutout means for cutting out the image data of the entry frame. Further, the character reading device includes a character recognition unit that recognizes a character or a symbol corresponding to the image data extracted by the extraction unit with reference to the character dictionary, and the first character recognition unit that is recognized by the character recognition unit. Selecting means for outputting a selection signal for identifying a word in the second entry frame based on the characters or symbols in the entry frame, and referring to a specific word set in the word dictionary based on the selection signal. And word identification means for identifying a word in the second entry frame recognized by the character recognition means from the word set.

According to the second invention, the following operation is performed. First, from the image information stored in the storage means, image data corresponding to the first entry frame of the form is cut out by the cutout means and provided to the character recognition means. The character recognition means refers to the character dictionary and recognizes a character or a symbol corresponding to the image data of the first entry frame. The recognition result of the character recognition means is given to the selection means, and the selection means outputs a selection signal for selecting a word in the second entry frame. Next, image data corresponding to the second entry frame of the form is cut out from the image information stored in the storage means and provided to the character recognition means. The character recognition means refers to the character dictionary, recognizes a character or a symbol corresponding to the image data of the second entry frame, and gives the recognition result to the word identification means. The word identifying means identifies a word corresponding to the image data of the second entry frame from the word set in the word dictionary selected by the selection signal for selecting a word of the second entry frame.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS First Embodiment FIG. 1 is a block diagram of a character reading apparatus according to a first embodiment of the present invention. Elements common to those in FIG. Is attached. As shown in FIG. 3, for example, this character reading device stores an image scanner 11 that optically reads image information of a form 1 in a predetermined format, and stores image information of the form 1 read by the image scanner 11. Storage means (for example, image memory) 1
Two. In addition, the character reading device includes a format registration unit 13 in which the positions of the mark frame 2a and the character frames 2b to 2g of the form 1 for writing characters to be read and the character types to be written therein are previously registered. Have. The image memory 12 has a format registration unit 13
Mark 2a and character frame 2b of the form 1 based on the information of
Cutting means for cutting out image data of ~ 2 g (for example,
A character cutout section 14 is connected. On the output side of the character extracting unit 14, a character recognizing unit (for example, a character recognizing unit) 15
Is connected.

The character recognition section 15 refers to a specific character set in the character dictionary 16A and recognizes a character or a symbol corresponding to the image data provided from the character extraction section 14. The character dictionary 16A has a plurality of character sets 16a, 16b, 16c,..., 16z for distinguishing and recognizing characters of a specific group. For example, the character set 16a is for registering characteristic data for distinguishing and recognizing the numbers "0" and "1", and the character set 16b is for registering and distinguishing the numbers "0" to "5". This is the registered feature data. The character set 16c registers feature data for distinguishing and recognizing the numbers “0” to “6”. The character set 16z recognizes and distinguishes all the numbers “0” to “9”. Is registered.

On the output side of the character recognizing unit 15, a character code as a result of recognition is output, and a selecting means (for example, a character set selecting unit) 17 is connected. The character set selection unit 17 performs the second entry based on the information from the format registration unit 13 and the recognition result of the symbol in the first entry box (for example, mark box) 2 a output from the character recognition unit 15. It outputs a selection signal SEL for recognizing image data of a frame (for example, a character frame) 2b. The selection signal SEL output from the character set selection unit 17 is transmitted to the character dictionary 16A.
To the character recognition unit 15 from the character dictionary 16A.
Is provided with the selected character set 16a and the like.

FIG. 4 is a flowchart showing the operation of FIG. Next, the operation of the character reading apparatus of FIG. 1 will be described with reference to FIG. 4 for the case of reading the form 1 of FIG. As shown in FIG. 3, a mark and a character are entered at a predetermined position in the entry column 2 of the form 1, and are inputted to the image scanner 11 of FIG. In step S1 of FIG.
Is read and stored in the image memory 12. In step S <b> 2, the character extracting unit 14 operates, and the image data of the mark frame 2 a is extracted from the image memory 12 based on the information of the format registration unit 13 and provided to the character recognition unit 15. Also, the format registration unit 13 sends a character set selection unit 17 a
Information indicating that there are four mark positions corresponding to Meiji, Taisho, Showa and Heisei is given to the mark frame 2a.

In step S3, it is determined whether the extracted image data is a mark or a character. in this case,
Since it is a mark, the process proceeds to step S4. Step S4
, The mark detection selection signal SEL is
6A, the character recognition unit 1
5, the characteristic data for mark identification is output. The character recognition unit 15 detects the position of the mark based on the feature data provided from the character dictionary 16A. After the detection of the mark position, the process proceeds to step S5. In step S5, a mark is recognized based on the mark position. In this case, the year corresponding to the mark is recognized as “Taisho”, and for example, a code “2” corresponding to “Taisho” is output. Furthermore, the mark frame 2a output from the character recognition unit 15
Is given to the character set selection unit 17, and the process proceeds to step S6.

In step S6, the recognition result of the mark frame 2a is output to, for example, a host computer (not shown), and the recognition process of the mark frame 2a ends. After step S6, the process proceeds to step S7, where it is determined whether all character recognition processes have been completed. If the character recognition process remains, the process returns to step S2, and if all the processes have been completed, the character reading process of the form 1 ends.
In this case, since the processing of the character frames 2b to 2g remains,
Returning to step S2, the character extracting unit 14 stores the image memory 12 based on the information in the format registering unit 13.
, The image data of the character frame 2b is cut out and provided to the character recognition unit 15A. After step S2, the process proceeds to step S3. In step S3, it is determined whether the extracted image data is a mark or a character. in this case,
Since it is a character, the process proceeds to step S8.

In step S8, a selection signal SEL for recognizing image data of the character frame 2b is output from the character set selection unit 17 to the character dictionary 16A. In this case, since the recognition result of the mark frame 2a is "Taisho", the selection signal SEL for selecting the character set 16a is given to the character dictionary 16A. Thereby, the character dictionary 16A
The character set 16a composed of feature data for distinguishing and recognizing the numbers "0" and "1" is provided to the character recognition unit 15 for character frame 2b recognition. After step S8, the process proceeds to step S9. In step S9, the character recognition unit 15 performs character recognition of the character frame 2b based on the character set 16a. And the character frame 2b
Is recognized as, for example, a numeral "1", and a code "1" corresponding to the numeral "1" is output. The recognition result of the character frame 2b output from the character recognition unit 15 is
It is provided to the character set selection unit 17. After step S9, the process proceeds to step S6. In step S6, the recognition result of the mark frame 2a is output to the host computer,
The recognition process of the mark frame 2a ends.

When the recognition process of the character frame 2b is completed, the process returns to step S2, and the character extracting unit 14 stores the image memory 12 based on the information of the format registering unit 13.
, The image data of the character frame 2c is cut out and given to the character recognition unit 15. In step S8, since the character set selection unit 17 has given the numeral "1" as the recognition result of the character frame 2b, the character set 16b is used as the selection signal SEL for image data recognition of the character frame 2c.
Is supplied to the character dictionary 16A.
As a result, the character set 16 composed of the feature data for distinguishing and recognizing the numbers “0” to “5” for the character frame 2 c from the character dictionary 16 A to the character recognition unit 15.
6b is provided. Thus, in step S9, the character recognition unit 15 performs character recognition of the character frame 2b based on the character set 16b. Then, the character frame 2c is recognized as, for example, a numeral "2", and a code "2" corresponding to the numeral "2" is output. The recognition result of the character frame 2c output from the character recognition unit 15A is output to the character set selection unit 17
Given to.

Similarly, the image data of the character frames 2d to 2g are sequentially recognized in accordance with the character set 16a and the like selected by the selection signal SEL given according to the recognition result of the previous character frames 2c to 2f. . Here, if the recognition result of the mark frame 2a is “Showa”, the code “3” corresponding to “Showa” is output from the character recognition unit 15 in step S5, and given to the character set selection unit 17. Then, in step S8, a signal for selecting the character set 16c is given from the character set selection unit 17 to the character dictionary 16A as a selection signal SEL for recognizing image data of the character frame 2b. As a result, the character dictionary 16A sends the character
For recognition, a character set 16c composed of feature data for distinguishing and recognizing numbers "0" to "6" is provided.

As described above, in the first embodiment,
When referring to the character dictionary 16A for recognizing the image data of each of the character frames 2b to 2g, the character sets 16a to 16z selected according to the recognition results of the mark frames 2a and the character frames 2b to 2f are used. . This allows
Since the character candidates to be recognized can be limited, there is an advantage that the time for the recognition process can be reduced.
Furthermore, since an invalid character is not erroneously selected, there is an advantage that the recognition rate can be improved.

Second Embodiment FIG. 5 is a block diagram of a character reading apparatus according to a second embodiment of the present invention. Elements common to those in FIG. 1 are denoted by the same reference numerals. . The character reading device shown in FIG. 5 includes an image scanner 11, an image memory 12, a format registration unit 13, and a character extraction unit 1 similar to the character reading device shown in FIG.
4 and a character recognition unit 15. On the other hand, this character reading device has a character dictionary 16 similar to the conventional one.
That is, the character dictionary 16 has registered therein the characteristics of the character patterns necessary for character recognition, and outputs the characteristic data of the necessary character type to the character recognition unit 15 according to the information provided from the format registration unit 13. Things. A selection means (for example, a word set selection unit) 18 is connected to the output side of the character recognition unit 15. Word set selector 18
Outputs a selection signal SEL for word recognition of the second entry frame based on the recognition result of the first entry frame output from the character recognition unit 15. The selection signal SEL output from the word set selection unit 17 is provided to the word dictionary 19.

The word dictionary 19 includes a plurality of word sets 19 for distinguishing and recognizing words of a specific group.
a, 19b,... For example, the word set 19a
, "AB" to distinguish and recognize multiple banks
., "BCD", "CDE",..., And the word set 19b contains "XYZ", "YZA", "Z
AX ",... Are registered. Then, according to the selection signal SEL given from the word set selection unit 17, the word dictionary stores the designated word set 19a.
Are output to the word identification means (for example, a word identification unit) 20. The word identification unit 20 identifies a word composed of a character string recognized by the character identification unit 15 from a word set 19 a or the like provided from the word dictionary 19 and outputs the word.

FIG. 6 is a diagram showing another part of the form 1 to be read. This form 1 has character frames 3a to 3f for entering names of financial institutions and the like, and character frames 4 for entering codes of the types of the financial institutions (for example, 1: bank, 2: credit union). Is provided. FIG. 7 is an explanatory diagram of the operations of the word set selection unit 18, the word dictionary 19, and the word recognition unit 20 in the character reading device of FIG. Next, with reference to FIG. 7, the operation of the character reading apparatus of FIG. 5 will be described for the case of reading the form 1 of FIG.

The image data of the character frames 3a to 3f of the form 1 is recognized in the character recognition unit 15 by referring to the character dictionary 16. Then, for example, as a recognition result of the character frame 3a, the first candidate “A”, the second candidate “P”, and the third candidate “D” are output in order of likelihood. In addition, a first candidate “B” and a second candidate “E” are output as a recognition result of the character frame 3b. Furthermore, as a recognition result of the character frame 3c, the first candidate “O”, the second candidate “C”, and the third candidate “D”
Is output. Character frames 3d to 3f are recognized as blank. On the other hand, the recognition result of the character frame 4 is “1”, that is, “bank”.
Is recognized. These recognition results are provided to the word set selection unit 18 and the word recognition unit 20. In the word set selection unit 18, based on the given recognition result of the character frame 4,
A selection signal SEL for selecting the word set 19 a in which the bank name is registered is output to the word dictionary 19. As a result, a word set 19a in which a plurality of bank names are registered is provided from the word dictionary 19 to the word recognition unit 20.

In the word recognition unit 20, a plurality of words are generated by combining the recognition results of the character frames 3a to 3c in which the characters have been recognized in descending order of certainty, and a plurality of words are generated. It is determined whether or not the words are registered in the word set 19a in descending order of probability.
Then, the word that has been confirmed to be registered first is identified as being correctly read, and the word is recognized by the word recognition unit 2.
Output from 0. As described above, the character reading apparatus according to the second embodiment is based on the word dictionary 19 in which a plurality of reading candidate words are registered in advance as a word set, and the character frame 4 based on the recognition result of the character frame 4 of the form 1. A word set selection unit 18 for selecting a word set 19a or the like including the words in 3a to 3f, and whether or not the word is registered in the selected word set 19a or the like based on the likelihood of the recognized character And a word identification unit 20 for determining For this reason, there is an advantage that a reading error such as outputting an impossible word as a recognition result can be prevented.

The present invention is not limited to the above embodiment, and various modifications are possible. For example, there are the following modifications (a) to (e). (A) Instead of the mark frame 2a of the form 1 in FIG.
For example, a character frame for entering the year by code may be provided.
Further, instead of the character frame 4 of the form 1 in FIG. 6, for example, a mark frame for specifying the type by a mark may be provided. (B) In the character reading apparatus of FIG. 1, the character sets 16a and the like are used for reading the character frames 2b, 2c,... Of the form 1 in FIG. One character frame 2b, 2c may be used in common, and a character set such as two-digit numbers “01” to “15”, “01” to “64” may be used.

(C) The character set selection section 17 in the character reading apparatus of FIG. 1 outputs the selection signal SEL based on the recognition result of one mark frame 2a, one character frame 2b, etc. One or more mark frames 2a and character frames 2b
The selection signal SEL may be output based on a recognition result such as the above. As a result, it is possible to select a character set in more detail, and it is possible to further reduce the time required for the recognition process and improve the recognition rate. (D) On the output side of the character recognition unit 15 in FIG. 1, the word set selection unit 18, the word dictionary 19, and the word recognition unit 20 in FIG.
May be added. Thereby, the combined advantages of the first and second embodiments can be obtained. (E) The character reading device shown in FIGS.
4. Character Recognition Unit The character set selection unit 17, the word set selection unit 18, the word identification unit 20, and the like are configured as individual processing units, but these processes are performed under program control using a computer. Is also good.

[0028]

As described above in detail, according to the first aspect, the selection for designating the character set to be used for recognizing the second entry box based on the recognition result of the first entry box. Means, and a character dictionary for outputting a specific character set in response to a selection signal from the selection means. This makes it possible to reduce the number of characters to be recognized, thereby shortening the recognition processing time and improving the recognition rate. Second
According to the invention of the above, based on the recognition result of the first entry frame,
It has a selecting means for designating a word set used for identifying a word entered in the second entry box, and a word dictionary for outputting a specific word set in response to a selection signal from the selecting means. As a result, a meaningless combination of characters is not output as a recognition result, and the recognition rate can be improved.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a character reading device according to a first embodiment of the present invention.

FIG. 2 is a configuration diagram of a conventional character reading system.

FIG. 3 is a diagram showing a part of a form 1 to be read;

FIG. 4 is a flowchart showing the operation of FIG.

FIG. 5 is a configuration diagram of a character reading device according to a second embodiment of the present invention.

FIG. 6 is a diagram showing another part of the form 1 to be read.

FIG. 7 is an explanatory diagram of operations of a word set selecting unit, a word dictionary, and a word recognizing unit in the character reading apparatus of FIG.

[Explanation of symbols]

 1 Form 2a Mark frame 2b-2g, 3a-3f, 4 character frame 11 Image scanner 12 Image memory 13 Format registration unit 14 Character extraction unit 15 Character recognition unit 16, 16A Character dictionary 16a-16z Character set 17 Character set selection unit 18 Word set selection unit 19 Word dictionary 19a, 19b Word set 20 Word identification unit

Claims (2)

[Claims] 1. A first entry box for entering a character or symbol, and a second entry for entering a character in a character set corresponding to the character or symbol entered in the first entry box. In a character reading device that reads a form having a frame and recognizes a character or a symbol written on the form, a character in which characteristic data for recognizing a character or a symbol to be read is registered for each character set. Dictionary, storage means for storing the image information of the form, cutout means for cutting out the image data of the first and second entry frames from the image information stored in the storage means, based on a selection signal A character recognition unit that refers to a specific character set in the selected character dictionary and recognizes a character or a symbol corresponding to the image data extracted by the extraction unit; Selecting means for outputting the selection signal for recognizing image data of the second entry frame based on characters or symbols in the first entry frame recognized by the recognition means. Character reading device. 2. A first entry box for entering a character or symbol, and a second entry for entering a word in a word set corresponding to the character or symbol entered in the first entry box. A character reading device that reads a form having a frame and recognizes a character, symbol, or word written on the form includes a character dictionary in which feature data for identifying a character or a symbol to be read is registered. A word dictionary in which a plurality of words to be read are registered for each word set; storage means for storing the image information of the form; and the first and second images from the image information stored in the storage means. And a character recognition unit that recognizes a character or a symbol corresponding to the image data extracted by the extraction unit with reference to the character dictionary. A step, a selection unit that outputs a selection signal for word identification of the second entry box based on a character or a symbol in the first entry box recognized by the character recognition unit; And referring to a specific word set in the word dictionary on the basis of the second
And a word identifying means for identifying a word in the entry frame from the word set.