JP2002109468A - Character recognition device and character recognition method - Google Patents

Abstract

(57) [Problem] To provide a character reading device and a character reading method capable of accurately identifying an existing form created without considering form identification and character recognition. With the goal. SOLUTION: A plurality of form definition information for identifying a form and format control information for recognizing a character recorded in a form corresponding to each of the plurality of form definition information are stored in advance and input. The form for character recognition is identified as a specific form based on the form image of the form for character recognition and the form definition information, and the character is recognized using the corresponding format control information based on the identification result.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition device for recognizing characters recorded on a form (hereinafter referred to as OCR).
More particularly, the present invention relates to a character recognizing apparatus and a character recognizing method capable of easily recognizing characters by identifying the type of an existing form not designed for character recognition.

[0002]

2. Description of the Related Art Conventionally, in general, when it is desired to read characters recorded on a plurality of types of forms in OCR,
The form had to be designed for OCR read-only so that the OCR could identify the form. In a form designed only for OCR reading, it is necessary to record (print) an ID number in a predetermined position in advance for form identification.
In OCR, format information (format information or format control information (F
C)) is stored in advance, and the type (ID
The target form was read using the format information corresponding to the number. The format information includes the position of the field where the character to be read is recorded on the form, the number of digits of the read character, the character pitch, the character type,
This is information that specifies the type of character recognition dictionary used for character reading.

On the other hand, in recent years, there has been an increasing demand for reading characters recorded on existing forms that are not specifically designed for OCR reading. As described above, when reading characters recorded in an existing form, registering format information in the OCR, manually classifying a plurality of types of existing forms by type, and bundling them one by one. It was necessary to perform character reading by designating format information for each bundle (one batch).

Further, when reading characters recorded on a form by going one step further, coordinate information and format information of character entry frames (ruled lines) on the form are registered in the OCR in advance. The coordinate information of the ruled line is acquired from the image of the form input to the OCR, the form is identified by comparing the coordinate information of the ruled line of each form registered in advance, and the character is determined based on the format information corresponding to the identified form. Performing a read was done. According to this method, it is not necessary to manually sort the forms for each type before the reading process and to process the same form in one batch during the reading process. Therefore, reading processing can be performed by mixing a plurality of types of forms.

[0005] However, in this method, if the input form image is blurred, the ruled line cannot always be accurately extracted, so that there is a problem that the form cannot be correctly identified. In addition, there is a problem that even if the forms are of different types, the forms cannot be identified if the positions of the ruled lines are exactly the same.

[0006]

SUMMARY OF THE INVENTION The present invention has been made to solve such a problem, and it is an object of the present invention to accurately identify an existing form created without considering form identification and to perform character recognition. It is an object of the present invention to provide a character recognition device and a character recognition method capable of performing the following.

[0007]

According to the present invention, there is provided a form definition information holding means for holding a plurality of form definition information for identifying a form, and a character recorded on a form corresponding to each of the plurality of form definition information. Format control information holding means for holding format control information for recognizing a document, image input means for inputting a form image of a form for character recognition, a form image for inputting the form for character recognition, and the form definition Form identification means for identifying a specific form based on the information, and based on the form identification result of the form identification means, select format control information corresponding to the identified form from the format control information holding means, The input form image using the selected format control information Characterized by comprising a character recognizing section for recognizing characters.

According to such a configuration, it is possible to realize a character recognition device and a character recognition method capable of accurately identifying an existing form created without considering form identification and performing character recognition.

[0009]

Embodiments of the present invention will be described below in detail with reference to the drawings. FIG. 1 shows the OCR of the present invention.
1 is a diagram showing a system configuration of FIG. Optical character reader 1
Reference numeral 0 denotes a color scanner 11 and a personal computer 12. In the personal computer 12, a character recognition program 13 which is OCR software for realizing character recognition, and a form definition program 14 for setting form identification information for identifying a form are installed. The color scanner 10 has a function of photoelectrically converting a form image and outputting image information of three colors of RGB. The color scanner 10 has an automatic paper feeding function, takes out forms one by one from a hopper on which forms to be read are stacked, feeds the sheets, and feeds the sheets to a stacker via a reading position. The color scanner 10 reads the form image by photoelectrically converting the image on the form by the CCD color image sensor at the reading position of the fed form and reads the form image.
Output color image information. The color scanner 10 is provided with a normal stacker for discharging a form in which all characters can be correctly recognized and a reject stacker for discharging a form having unreadable characters (reject characters). Based on the result of character recognition, the fed form is copied to the normal stacker /
Discharge to one of the reject stackers. The magnetic disk device 15 is provided as an external storage device of the personal computer 12. Form definition information 16 and format control information 17 are stored and registered in the magnetic disk device 15.

In this embodiment, a software character recognition type OCR having a color scanner 11 and a personal computer 12 will be described. However, the present invention is not limited to this, and the color scanner may be integrated, and may be a dedicated machine having a character recognition function and a form definition function for setting form identification information for identifying the form. In the case of a special-purpose machine, a sorter for classifying the form after the character recognition processing by type can be connected. The sorter has multiple output pockets,
By discharging the identified forms into the paper discharge pockets according to their types, the forms are classified.

A form identification item defined as identification information for identifying a form includes a form size, a corner cut, a binding hole, and a preprint (preprint) as shown in FIG.

The form size is information defined by the height and width of the form. The corner cut is the position (upper right, lower right, upper left, lower left) and size (width, width, etc.) of the corner cut provided in at least one of the four corners of the form.
Height). FIG. 3 illustrates a form in which corner cuts are provided at the upper left and lower right. The binding hole is information defined by the position of the binding hole (coordinates having the origin at the upper left of the form) provided in the form for filing and the diameter of the hole. FIG. 4 illustrates a form having a binding hole H1 of diameter R1 provided at the position of coordinates (X1, Y1) and a binding hole H2 of diameter R2 provided at the position of coordinates (X1, Y2). Preprinting (preprinting) is information indicating an area of a character string that has been printed (preprinted) on a form in advance, and includes a position (coordinates with the origin at the upper left of the form), width, height,
Information defined by a character string, character size (point number), font (character type), character pitch, and color existing in the area. The preprint (preprint) is not limited to characters, but may be a graphic image. FIG. 5 shows a form having a preprint area L including a character string preprinted (preprinted) as “@ Corporation”. This pre-printing area L has coordinates (X3,
The form identification item, which is provided at the position Y3) and is used to identify the form, is defined by executing the form definition program 14.

When the form definition program 14 is executed, image information of an unfilled form to be identified is read by the color scanner 10 and displayed on a display device (not shown) of the personal computer 12.
The user sets the definition information by designating the position and area with the mouse and inputting information with the keyboard while looking at the displayed form image. The display screen displays the item names of the form identification items and the input fields for the corresponding definition information.

In response to a request for setting a form identification item from a program or a designation of an item name by a user using a mouse, the user can use a mouse to specify a corresponding part of the displayed form image corresponding to the item name for which the definition information is set. Is specified as a rectangle. Within the range specified by the rectangle, the program searches for an image corresponding to the setting item requested to be set or specified by the user, and acquires information to be set. For example, when the form size is set as a setting item and the entire form is designated by a rectangle, the outer shape of the form image is searched, the number of dots of the image is counted, and the counted number of dots and the photoelectric conversion of the scanner are used. Calculate from the resolution and the vertical and horizontal length of the form,
Displayed in the definition information input field.

As for other setting items, a target image is searched for in a range designated by a rectangle, the number of dots is counted, and coordinates having the origin at the upper left position of the form image are obtained from: Ask for the setting information.
Note that information on the character string in the blur print area and the font, pitch, size, and color of each character constituting the character string can be obtained by character recognition in the range specified by the rectangle. As for the color of the character, character recognition is performed for each of the three RGB image information output from the color scanner 10, and the color is determined based on which color image or the like was used for character recognition.

When a character can be recognized from the R and G images, it can be determined that the character is a blue (system) character. Similarly, when a character can be recognized in the image of G and B, it is a red (system) character, and when a character can be recognized in the images of B and R, it is a green (system) character. Can be determined. Similarly, when a character can be recognized in the R, G, and B images, it can be determined that the character is a black (system) character. The form identification item thus defined is stored and registered in the magnetic disk device 15 as the form definition information 16.

On the other hand, the format control information 17 of each form in which the form definition information 16 is registered is also stored and registered in the magnetic disk device 15. The form definition information 16 is associated with the format control information 17 so that the format control information 17 can be called from the form definition information 16. The format control information 17 is set in advance for each form, and is stored in the magnetic disk device 15 in advance as a file. This format control information 17
For the setting method, the conventional technique may be used, and the detailed description is omitted. Further, the form definition program 14 may be configured so that the format information 17 is also set together with the form identification items along with the execution of the form definition program 14.

Hereinafter, the character recognition operation will be described. As a prerequisite for character recognition, it is assumed that the form definition information 16 corresponding to the existing form to be read and the format control information 17 corresponding thereto are registered in the magnetic disk device 15 in advance as described above.

First, a form to be read is read by the color scanner 10 and form image information is output to the personal computer 12. Personal computer 1
In step 2, the character recognition program 13 starts processing and starts a character recognition operation. First, the character recognition program 13
The form type is identified based on the form image input from the color scanner 10. The operation of this form identification process will be described with reference to the flowcharts shown in FIGS.

In FIG. 6, first, a variable I is set to a numerical value 1 (step S61). Next, the I-th form definition information is read from the magnetic disk device 15 (step S6).
2). Subsequently, in step S63, form identification is performed to determine whether the form having the input form image is the same as the form having the I-th form definition information.

The details of this form identification are shown in the flowchart of FIG. First, a numerical value “0” is set to the variable R (step S70). In a succeeding step S71, an image obtained by rotating the input form image by R degrees is obtained. Since the numerical value “0” is currently set in the variable R, an image of 0 degrees, that is, a form image that is input without any rotation processing of the input form image is obtained. In subsequent steps S72 to S75, a comparison is made as to whether or not the form image obtained in step S71 matches each form identification item defined in the I-th form definition information obtained in step S61.

In step S72, the form sizes are compared. Search the input form image to find the height and width of the form. The length and width of the input form obtained and I
Compare with the form size defined in the second form definition information. If the result of this comparison is a match, step S7
Proceed to 3. If they do not match, the process proceeds to step S77. If the form size is not defined in the I-th form definition information, the process proceeds to step S73.

In step S73, a comparison is made between corner cuts. The input form image is searched to determine whether there is a corner cut of the defined size (width and height) at the position defined by the I-th form definition information. If the result of this comparison is a match, step S
Proceed to 74. If they do not match, the process proceeds to step S77. If the corner cut is not defined in the I-th form definition information, the process proceeds to step S74.

In step S74, the binding holes are compared. The input form image is searched to determine whether or not a binding hole having the defined diameter exists at a position defined by the I-th form definition information. If the result of this comparison is a match, the flow proceeds to step S75. If they do not match, the process proceeds to step S77. If the corner cut is not defined in the I-th form definition information,
Proceed to step S75.

In step S75, pre-print comparison is performed. The input form image is searched for and is composed of characters having a size (number of points), a font (character type), a character pitch, and a color defined in the preprint area defined by the I-th form definition information. A comparison is made to determine whether the defined character string exists. This comparison is performed as follows. First, character recognition is performed on the assumption that a character string having a size (number of points), font (character type), character pitch, and color defined in the preprint area defined in the I-th form definition information. . As a result of this character recognition, if the obtained character string is the same as the above-defined character string, it means that the comparison judgment has been made. If the result of this comparison is a match, the flow proceeds to step S76. If they do not match, the process proceeds to step S77. If the pre-print area is not defined in the I-th form definition information,
Proceed to step S76.

In step S76, the input form image matches the form identification item defined in the I-th form definition information in all the comparison determinations from step S72 to step S75.
It is identified as the form defined in the second form definition information. Thereafter, the process proceeds to step S63 shown in FIG.

On the other hand, the input form image is
If it is determined that the values do not match in any one of the comparison determinations from 72 to step S75, the variable R is updated by adding a numerical value “90” to the variable R in step S77.
It is determined whether or not the updated variable R is equal to or exceeds the numerical value "360" (step S78). In this determination, in all cases of the input form image rotated by 0 degrees, the image rotated by 90 degrees, the image rotated by 180 degrees, and the image rotated by 270 degrees, the above-described steps S72 to S75 are performed. It is determined whether the comparison has been completed.

This determination is made because if the form to be read is loaded on the hopper of the color scanner 10 and the upper end of the form is correctly loaded with the paper feed roller facing, the input image is rotated. It is sufficient to perform the form identification processing without doing it, but it may not always be loaded at the correct position. This is because the form identification process is performed on an image obtained by rotating the image by 90 degrees, 180 degrees, and 270 degrees. Thus, the form identification process can be performed correctly regardless of the direction of loading of the form with respect to the hopper of the color scanner 10.

If the result of determination in step S78 is that variable R is less than or equal to numerical value "360", the flow returns to step S71, and
Hereinafter, similarly, the processing from step S71 to step S77 is performed. If the variable R is equal to or larger than the numerical value "360", the process proceeds to step S63 shown in FIG.

In step S63, the variable I is set to a numerical value "1".
To update the variable I. In a succeeding step S64, it is determined whether or not the numerical value of the variable I has exceeded the number of definition information registered in the form definition information 16. This decision
It is determined whether or not form identification has been performed on the input form image using all the form definition information registered in the form definition information 16. If the comparison with all the form definition information has not been completed, the process returns to step S61, and the processes from step S61 to step S63 are performed in the same manner. If it is determined in step S64 that the form identification using all the form definition information has been completed, the process proceeds to step S65.

In step S65, it is determined whether the number of the form images input in the processing in step S62 that matches the form definition information registered in the form definition information 16 is one or more. If the number of matches is one, the form identification process ends normally. In this case, the form having the input form image is the form defined by the form definition information uniquely identified in step S76. If the number of matches is plural or zero, the form identification processing is abnormally terminated.

When the form definition information that matches the input form image is identified, the format control information associated with the matched form definition information is read from the format control information 17. The character recognition program 13 executes character recognition from the input form image using the read format control information. Optical character reader 1
In the case where a sorter is connected to 0, the form after the character recognition processing is discharged by a sorter to a plurality of discharge pockets for each type, thereby enabling sorting.

One example of a large effect that can be expected by implementing the present invention is a character reading process of a payment notice of a fixed asset tax, a city planning tax, a water bill, etc. issued by a local government in a financial institution. These notification forms are forms designed and created by the respective municipalities or the like to perform OCR processing in their own information processing systems. However, from the viewpoint of the financial institution's OCR processing, these forms are merely existing forms. Therefore, when a request from a customer for a transfer process of various charges based on these notification sheets is received, these hundreds of existing forms are manually classified according to the issuing organization and the type of the notification form. At the same time, it was necessary to calculate the total transfer amount for each of these classified notices.

By applying the present invention to such processing, it is possible to easily identify hundreds of forms originally designed respectively. Can be easily realized, and the automatic calculation of the total amount for each notice can be realized. Moreover, if a sorter is provided in the OCR, the forms on which the character recognition processing has been completed can be classified according to their types. Therefore,
The calculation of the total amount and the classification by form of the form, which were conventionally performed manually by a person, can all be performed by machine processing, and the processing efficiency can be greatly improved.

[0035]

As described above, according to the present invention, it is possible to accurately identify an existing form created without considering form identification and recognize characters.

[Brief description of the drawings]

FIG. 1 is a diagram showing an OCR system configuration according to an embodiment of the present invention.

FIG. 2 is a diagram for explaining a form definition item that is form definition information.

FIG. 3 is a diagram for explaining a corner cut which is a form definition item.

FIG. 4 is a diagram for explaining a binding hole which is a form definition item.

FIG. 5 is a diagram for explaining preprints, which are form definition items.

FIG. 6 is a flowchart for explaining a form identification process.

FIG. 7 is a flowchart for explaining a form identification process in the flowchart of FIG. 6;

[Explanation of symbols]

 10 Optical character reader 11 Color scanner 12 Personal computer 13 Character recognition program 14 Form definition program 15 Magnetic disk unit 16 Form definition information 17 Format control information

Claims (7)

[Claims] 1. A form definition information holding means for holding a plurality of form definition information for identifying a form, and format control information for recognizing a character recorded on a form corresponding to each of the plurality of form definition information. Format control information holding means for holding a form, an image input means for inputting a form image of a form for character recognition, and a specific form based on the input form image and the form definition information for the form for character recognition. Form identification means for identifying as, based on the form identification result of this form identification means, select format control information corresponding to the identified form from the format control information holding means,
A character recognition unit for recognizing characters of the input form image using the selected format control information. 2. The form definition information includes a form size,
2. The method according to claim 1, wherein at least two of the corner cut position and size, the binding hole position and size, and the preprint information recorded in the specific area are selected and adopted. Character recognition device. 3. The form identification means, while inputting the input form image, outputs an image rotated 90 degrees,
4. The method according to claim 1, wherein four images, that is, an image rotated by 80 degrees and an image rotated by 270 degrees, are identified as specific forms based on the plurality of form definition information held in the form definition information holding unit. 2. The character recognition device according to claim 1, wherein: 4. A paper discharging means having a plurality of paper discharging pockets, and discharging the form in which the character is recognized to a specific paper discharging pocket for each type based on the form identification result of the form identifying means to sort the form. The character recognition device according to claim 1, further comprising: 5. A step of holding a plurality of form definition information for identifying a form, and a step of holding format control information for recognizing a character recorded on a form corresponding to each of the plurality of form definition information. Inputting a form image of a form for character recognition; identifying the form for character recognition as a specific form based on the input form image and the plurality of form definition information; and Selecting format control information corresponding to the identified form from the plurality of format control information based on the identification result, and recognizing characters of the input form image using the selected format control information. Character recognition method characterized by becoming. 6. The form identification step includes the step of defining the plurality of form definitions for four images of an image rotated by 90 degrees, an image rotated by 180 degrees, and an image rotated by 270 degrees while the input form image is input. 7. The character recognition method according to claim 6, wherein the form is identified as a specific form based on the information. 7. A method according to claim 7, further comprising the step of: providing a plurality of paper discharge pockets, discharging the character-recognized form to a specific discharge pocket for each type based on the form identification result, and classifying the form. 7. The character recognition device according to claim 6, wherein: