JPH0962780A - Reading format position correction method for optical character reader - Google Patents

Abstract

PROBLEM TO BE SOLVED: To read the characters even when the reading area of a business form is shifted from the standard format information. SOLUTION: The registered standard format information is registered in a storage 13. A format register device fetches the format image of an object business form into a work memory 12-1 of a personal computer 12 via an image reader 11. The format register device also reads out the registered format data corresponding to the object business form and sends them to the memory 12-1. Then a peripheral area surrounding the reading position of the format image is set around the reading position of an OCR that is extracted out of the format image, and a normal OCR reading position is secured for the object business form by the projection performed in the peripheral area. Then the registered format information is displaced with the format information corresponding to the normal reading position, so that the position to be read is corrected.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a reading format position of an optical character reading device for correcting a reading position when reading an item described in a form with an optical character reading device (hereinafter referred to as OCR). The present invention relates to a correction method.

[0002]

2. Description of the Related Art FIG. 2 is a diagram for explaining a conventional OCR reading method. The OCR 3 refers to the format information file 2 in which the read information (hereinafter, referred to as format information) is registered for the form 1 in which the character to be read is written, and performs the reading process. Format information file 2 contains
Position and attributes of characters to be read on form 1
Data such as (printed characters, handwriting, ANK, kanji, etc.) is stored for each item. The position and attribute of the character image to be read in the photoelectrically converted image image are specified by the format information, and the OCR 3
Character recognition can be performed along with it. Therefore, the format information file 2 needs to have the format information corresponding to all the form types to be read by the OCR 3. Correspondence between the form and the format information is performed by a sheet ID provided at a common position of each form. That is, in the OCR 3, first, the sheet ID (usually a 2 to 5 digit unique number for the form) provided at a position common to all the forms is read. Next, the OCR 3 retrieves the format information of the target form from the format information file 2 in which a plurality of format information is stored based on this sheet ID, and reads based on the retrieved format information. When registering this format information, the position of each reading item in the form is measured with a ruler and the value and the attribute of the character to be read are stored in the format information file 2 using a tool such as a personal computer. It is created by storing.

[0003]

However, the conventional OCR reading method has the following problems. In a reading method in which one type of format information is required for one type of form, for example, there is a case where a standard format of the form is determined nationwide. However, in some cases, each institution that uses a form or a maker that provides each institution with a device that creates the form creates its own form. In such a case, the reading position of the OCR 3 is slightly different and a reading failure occurs. In order to prevent reading errors, it was necessary to check the format information of the target documents to be read one by one and correct the deviation from the standard format. Moreover,
This correction may fail to be corrected because the difference between the form created in the standard format and the target form is visually detected. When the form is revised regularly, the revised data must be stored in the format information file 2 each time. That is, there is also a problem that a large amount of work is required.

[0004]

In order to solve the above problems, a first aspect of the present invention is a format information registration apparatus for registering format information for each form type used when reading an entry item in a form with an OCR. , The following OC
A read format position correction method for R is taken. That is, in the OCR read format position correction method of the first aspect of the invention, the OCR read position centered on the OCR read position captured from the format image of the target form registered in the format information registration device is acquired. Then, a peripheral area surrounding the reading position of the OCR on the format image is set. Then, the format image is projected in the X-axis direction and the Y-axis direction in the surrounding area to obtain the regular reading position in the target form, and the registered format information is associated with the regular reading position. The position to be read by the OCR is corrected by replacing it with the format information.

A second aspect of the present invention employs the following OCR read format position correction method in the OCR for reading the description items in the target form by using the format information for each registered form type. That is, the OCR of the second invention
In the reading format position correction method, the reading position on the format image is taken centering on the reading position of the OCR extracted from the registered format information for the target form by taking in the format image of the target form. Set the surrounding area. Then, projection of the format image in the X-axis direction and the Y-axis direction in the peripheral area is performed to obtain the regular reading position on the target form, and the regular reading position is replaced with the registered format information. By using the format information
The reading position of R is corrected.

A first aspect of the present invention relates to a format information registration device. According to the first aspect of the present invention, since the OCR read format position correction method is configured as described above, the format image of the target form is the format information. A peripheral area surrounding the reading position on the format image is set around the reading position of the OCR that is taken in by the registration device and extracted from the format information of the registered target form. Then, by performing projection in the X-axis direction and the Y-axis direction in the surrounding area on the captured format image, the regular reading position on the target form can be obtained. That is, even when the registered format information and the actual reading position are deviated, the actual reading position is accurately detected. By replacing the registered format information with the format information corresponding to the regular reading position, the position to be read by the OCR is corrected. The second invention is an invention for OCR. According to the second invention, the format image of the target form is captured in the OCR, and the OCR reading position extracted from the registered format information for the target form is centered. Thus, a peripheral area surrounding the reading position on the format image is set. Then, the format image is projected in the X-axis direction and the Y-axis direction in the surrounding area. This allows
A regular reading position on the target form is obtained. By using the format information corresponding to the regular reading position instead of the registered format information, the reading position of the OCR for the described item is corrected. Therefore, the above problem can be solved.

[0007]

BEST MODE FOR CARRYING OUT THE INVENTION First Embodiment FIG. 1 is a configuration block diagram of a format registration device showing a first embodiment of the present invention. The format registration device includes an image reader 11 that reads a format image of a form as an image, and a personal computer 12.
And a storage device 13 in which format information is registered. The personal computer 12 has a working memory 12-1, and the image of the format image from the image reader 11 is stored in the working memory 12-1.
It is designed to be input as data in. Next, the operation of the format registration device in FIG. 1 will be described. The image image of the format image of the form to be corrected is read by the image reader 11, and the data is photoelectrically converted and stored in the working memory 12-1 of the personal computer 12. The data of this image image is stored in the memory 12-1 in bit units. The blank part of the form is stored as "0", and the ruled line and character parts are stored as "1". Next, the personal computer 12
Reads out the format data corresponding to the form from the storage device 13 in which the format information is stored.

FIG. 3 is a diagram showing a part of the concept of format information. In FIG. 3, a reading area (field position) to be read, which is surrounded by ruled lines, is shown for each item. The reading area of item 1 has a dimension of Y ₁ from the top side of the form.
mm, dimension from left side is X ₁ mm, area height is ΔY ₁ mm,
And as the format data with the width of the area ΔX ₁ mm,
It is stored in the storage device 13. The reading area of item 2 is Y ₂ mm in size from the top side of the form, X ₂ mm in size from the left side, the height of the area is ΔY ₂ mm, and the width of the area is ΔX.
It is stored in the storage device 13 as ₂ mm format data. That is, it is assumed that these format data are registered in the storage device 13 as format information. Based on these format data, the corresponding reading area in the form image developed in the work memory 12-1 is specified. This identification method is
The size of the format data can be easily calculated by converting it into the number of dots in the memory according to the resolution of the image reader.

FIG. 4 is a diagram showing an image and format data of a form corresponding to the read item 1 in FIG. As shown in FIG. 4, when there is a gap between the reading area A on the form 20 indicated by the solid line frame and the area B specified by the format data indicated by the broken line frame, the OCR is correct as it is.
Cannot be read and format data must be corrected. FIG. 5 is a diagram showing a method of correcting the format data of FIG. FIG. 5 shows the actual reading area frame 21 of the form shown by the solid line and the position of the reading area frame 22 obtained from the format data shown by the broken line. First, an area around the broken line is set by α in the horizontal direction and β in the vertical direction. That is, the area 23 indicated by the dashed line is set. It is assumed that these α and β are obtained in advance so that the deviation of the area frame can be absorbed in the areas of all the items to be read in the form. Next, the work memory 12-1 is used to project the number of black dots in each of the X-axis and Y-axis directions within the area 23 indicated by a dashed line. The projection result 24 in the Y-axis direction is a count of black dots included from the upper side to the lower side in the area 23 of the Y-coordinate when the X-coordinate is fixed, and the projection result 25 in the X-axis direction is This is a count of black points included from the left side to the right side in the area 23 of the X coordinate when the Y coordinate is fixed.

In each of the projection results 24 and 25, there are two portions where the number of black spots is far greater than that of other areas. The portion where the number of black dots is by far the largest corresponds to the ruled line portion of the reading area 21 on the actual form. Since the position of the ruled line obtained from this projection is obtained by the number of dots from the upper and left sides of this form image, if this is calculated with the resolution of the image reader that acquired the data, the actual size of the form will be calculated as X. _{* 1} , Y _{* 1} , ΔX _{* 1} , ΔY _{* 1}
Is obtained. That is, each X _{* 1} , Y _{* 1} , ΔX _{* 1} , ΔY
_{* 1} is the value that is corrected according to the actual situation. Each obtained dimension _{X * 1, Y * 1,} ΔX * 1, ΔY * 1 a, as a new format data, each have been registered in the initial storage device 13 values _{X 1, Y 1, ΔX 1} , By replacing with ΔY ₁ , the format data according to the form is updated. By repeating this process for all the reading areas of the form, the format information can be updated. As described above, according to the second embodiment, based on the already registered format information, the working memory 12-1 is used for the image data of the target form, and the OCR format information that actually matches Is being corrected. Therefore, it is possible to perform the correction work without error and extremely efficiently. Further, in this embodiment, since the format information is corrected only by the format registration device, there is an advantage that the OCR main body is not affected at all.

Second Embodiment FIG. 6 is a block diagram of the OCR showing the second embodiment of the present invention. The OCR includes an image input unit 31 that reads the format image of the form as an image, a read control unit 32, and a storage device 33 in which the format information is registered. The read control unit 32 is provided with a working memory 32-1 and the image of the image in the format from the image input unit 31 is stored in the working memory 3-1.
2-1 is input as data. Next, the operation of the format registration device in FIG. 6 will be described. The image input unit 31 reads an image image of the form to be read, the data is photoelectrically converted, and the reading control unit 3
2 is stored in the work memory 32-1. The data of this image image is stored in the memory 32-1 in bit units. The blank part of the form is stored as "0", and the ruled line and character parts are stored as "1". Next, the reading control unit 32
Reads out the format data corresponding to the form from the storage device 33 in which the format information is stored.

FIG. 7 is a diagram showing a part of the concept of the format information in FIG. In FIG. 7, the reading area to be read surrounded by ruled lines is shown for each item. The reading area of item 1 is stored as format data in which the size from the top side of the form is Y ₁₁ mm, the size from the left side is X ₁₁ mm, the height of the area is ΔY ₁₁ mm, and the width of the area is ΔX ₁₁ mm. It is stored in the device 33. The reading area for item 2 is
The dimension from the top side of the form is Y ₁₂ mm, the dimension from the left side is X ₁₂
mm, the height of the area is ΔY ₁₂ mm, and the width of the area is ΔX ₁₂ mm, which is stored in the storage device 33. That is, it is assumed that these format data are registered in the storage device 13 as format information. From these format data, the working memory 3
The corresponding reading area in the form image developed in 2-1 is specified. This specifying method is easily specified by converting the size of the format data into the number of dots in the memory 32-1 according to the resolution of the image input unit.

FIG. 8 is a diagram showing an image of a form and format data corresponding to the read item 1 in FIG. As shown in FIG. 8, when there is a deviation between the reading area C on the form 40 indicated by the solid line frame and the area D specified by the format data indicated by the broken line frame, the OCR is correct as it is.
Cannot be read, and format data must be corrected. FIG. 9 is a diagram showing a method of correcting the format data according to FIG. FIG. 9 shows the actual reading area frame 41 of the form shown by the solid line and the position of the reading area frame 42 obtained from the format data shown by the broken line. First, an area around the broken line is set to be widened by α _{1 in} the horizontal direction and β _{1 in} the vertical direction. That is, the area 43 indicated by the dashed line is set. It is assumed that α ₁ and β ₁ are obtained in advance so that the shift of the area frame can be absorbed in the areas of all the items to be read in the form. Next, the work memory 32-1 is used to project the number of black points in each of the X-axis and Y-axis directions within the area 43 indicated by a dashed line. Projection result in Y-axis direction 44
Is a count of black dots included from the upper side to the lower side in the Y coordinate area 43 when the X coordinate is fixed, and the projection result 45 in the X axis direction is when the Y coordinate is fixed. The number of black dots included in the left side to the right side of the X-coordinate area 43 is counted.

In each of the projection results 44 and 45, there are two portions where the number of black spots is far greater than that of other areas. The portion where the number of black dots is by far the largest corresponds to the ruled line portion of the reading area 41 on the actual form. Since the position of the ruled line obtained from this projection is obtained by the number of dots from the upper and left sides of this form image, if this is calculated by the resolution of the image input unit 31 that acquired the data, it will be the actual size on the form. X _{* 11} , Y _{* 11} , ΔX
The values of _{* 11} and ΔY _{* 11} are obtained. That is, each X _{* 11} , Y
_{* 11} , ΔX _{* 11} , and ΔY _{* 11} are values that are actually corrected. Each of the obtained dimension values X _{* 11} , Y _{* 11} , Δ
X * _11, [Delta] Y * _11, as new format data, the values have been registered in the initial storage device 33 X _11,
By using the reading instead of Y ₁₁ , ΔX ₁₁ , and ΔY ₁₁ , the reading item can be read accurately. By repeating this process for all the reading areas of the form, it is possible to correctly read the OCR for all the reading areas.

As described above, according to the second embodiment,
By referring to the already-registered format information, the working memory 32-1 is used to correct the format information in accordance with the image data of the target form, and the OCR reads using the corrected format information. Therefore, even if there is a slight deviation in the actual reading area of the form, correct reading can be performed. In addition,
The present invention is not limited to the above embodiment, and various modifications can be made. For example, in the first embodiment, the dedicated image sensor 11 is used, but it has a function of capturing the image of the form in the OCR and a function of transmitting the captured image data to the personal computer 12. In that case, it can be used also as the OCR. In addition, in the second embodiment, it is premised that there is format data for each form type, but if the range of deviation in all forms is within the correction range, one form for all forms It is possible to perform correction by commonly using the format data of. In this case, the storage device 33 for registering the format information can be made extremely small.

[0016]

As described in detail above, according to the first aspect of the invention, in the format information registration device, the format image of the target form is taken in and the reading position extracted from the registered format information is centered, Set the surrounding area surrounding the reading position on the format image, find the regular reading position by projection in the surrounding area,
Since the registered format information is replaced with the format information corresponding to the regular reading position, the correction of the position to be read by the optical character reading device can be performed without error in the correction work and extremely efficiently. Therefore, even if there is a slight deviation in the actual reading area of the form, correct reading can be performed. Further, since the format information is corrected only by the format registration device, there is an advantage that the OCR body is not affected at all.
According to the second aspect of the invention, the format image of the target form is captured in the OCR, and the surrounding area surrounding the reading position on the format image is set around the reading position extracted from the registered format information. In order to correct the reading position for the described item by performing the projection of the inside, obtaining the regular reading position in the target form, and using the format information corresponding to the regular reading position instead of the registered format information. ,
Correct reading can be performed even if there is a slight deviation in the actual reading area of the form.

[Brief description of drawings]

FIG. 1 is a configuration block diagram of a format registration device showing a first embodiment of the present invention.

FIG. 2 is a diagram illustrating a conventional OCR reading method.

FIG. 3 is a diagram showing a part of the concept of format information.

FIG. 4 is a diagram showing an image and format data of a form corresponding to read item 1 in FIG.

5 is a diagram showing a method of correcting the format data of FIG.

FIG. 6 is a configuration block diagram of an OCR showing a second embodiment of the present invention.

7 is a diagram showing a part of the concept of format information in FIG.

FIG. 8 is a diagram showing an image and format data of a form corresponding to read item 1 in FIG.

9 is a diagram showing a method for correcting format data according to FIG.

[Explanation of symbols]

 11 Image Reader 12 Personal Computer 13,33 Storage Device 31 Image Input Unit 32 Reading Control Unit 20,40 Form

Claims (2)

[Claims] 1. A format information registration device for registering format information for each form type, which is used when an item to be written in a form is read by an optical character reading device, wherein a format image of a target form is taken in, and the format information registration device is used. Centering the reading position of the optical character reading device extracted from the format information of the target form registered in, the surrounding area surrounding the reading position of the optical character reading device on the format image is set, X in the surrounding area for the image
The optical character reading device by performing projection in the axial direction and the Y-axis direction to obtain the regular reading position in the target form and replacing the registered format information with the format information corresponding to the regular reading position. A method for correcting a read format position of an optical character reading device, which comprises correcting a position to be read by the device. 2. An optical character reading device for reading items described in a target form by using format information for each registered form type, wherein a format image of the target form is taken in and the registered format for the target form is read. A surrounding area surrounding the reading position on the format image is set around the reading position of the optical character reading device extracted from the information, and X in the surrounding area is set with respect to the format image.
By performing projection in the axial direction and the Y-axis direction to obtain the regular reading position in the target form, and using the format information corresponding to the regular reading position instead of the registered format information, A method for correcting a read format position of an optical character reading device, which comprises correcting the reading position of a character reading device.