JP4544703B2 - Form reading method and form reading system - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a form reading method and a form reading system for identifying the format of a form and determining the position of a processing target area such as character recognition in an image-based form processing system in which images of various forms are mixedly input.
[0002]
[Prior art]
In a so-called core business form image processing system that inputs image data of a wide variety of forms and recognizes characters in the character recognition target area or filing an image, first the image data of the input form is input. It is necessary to classify according to business.
[0003]
In the classification system, when image data is input from an image input device such as an image scanner or FAX, (1) the form is classified and sorted in advance at the paper stage by type, and an operator inputs the image. Method of specifying the type of form (form format specification), (2) Method of inputting an image by adding a form (batch form, etc.) for determining the type of form to the top of the bundle of the corresponding form (specifying batch form) (3) A method such as a method of attaching a unique ID (by character) to a form and identifying the type of the form from the character recognition result (sheet ID identification) has been used.
[0004]
In recent form image processing systems, not only forms created exclusively for the system but also existing forms and forms created by an unspecified number of outsiders have been required to be processed.
Therefore, as a new classification method, (4) the ruled line feature of the image data of the form to be classified is extracted, the feature is registered and classified in advance as the format of the image, and individual image data is input. A method of identifying the type of a form by comparing the characteristics of the ruled line of the image data with the characteristics of the ruled lines of a plurality of registered formats (ruled line characteristic identification), (5) a unique pattern of the form to be classified An area (such as a form title) having a pattern is registered as a form feature area, and the type of form is identified (area feature identification) in comparison with the pattern of each input image data.
[0005]
[Problems to be solved by the invention]
In the core industry form image processing system, it is necessary to also process forms created by an unspecified number of outsiders. In this case, form identification is often performed using ruled line features and area features as described above. .
However, depending on the system, if (B) A4 vertical size form and A4 horizontal size form, but the paper sizes are the same but different paper directions are subject to mixed input processing, As described above, in the horizontal direction, an image that cannot be acquired by an image input device (such as a scanner or a FAX) may be a target for mixed input processing.
[0006]
In the case of (A), images are often acquired in either the A4 horizontal direction or the A4 vertical direction in order to prevent skew by the image input device, shorten the image acquisition processing time, and simplify the operation.
In the case of (b), the B4 horizontal form is input in the B4 vertical direction.
In such a case, the forms are sorted in advance according to the direction, only the paper in the same direction is set in the image input device, the operator or application specifies the paper direction, and the image is rotated in the specified direction after the image is acquired. Thus, processing such as form identification and character recognition is performed.
[0007]
However, this requires manual sorting and cannot be said to be an original mixed input. Therefore, it is desired to be able to perform form identification and character recognition by inputting images in a mixed form with mixed paper orientations.
[0008]
[Means for Solving the Problems]
Therefore, the present invention reads the image data of a plurality of types of forms to be read, and the ruled line feature information in a state rotated from the forward direction so that the form sizes are aligned as form identifying feature information for identifying each form. It is registered in advance in the form format storage memory together with the rotation direction information, and the form information is identified by collating the feature information of the image data of the form inputted individually with the form registration feature information registered in advance. The image is rotated according to the rotation direction information of the form to be obtained to obtain an image in the forward direction, and processing target area information such as character recognition is set for the image in the forward direction.
[0009]
Furthermore, specifically, the image size of the form identification feature information extracted from the image input means for inputting the image data of the form to be read and storing the image data in the image memory and the image data of all the handled forms A form format storage memory in which ruled line feature information in a state rotated in the forward direction so as to be aligned is registered in advance together with the rotation direction information, a feature information extraction processing unit for extracting feature information of an input image, and extracted feature information Comprising a form identification means for identifying a form by collating with the previously registered form identification feature information, and an image rotation processing unit for rotating the input image. Rotating the image with information to obtain a forward image, and setting processing target area information such as character recognition for the forward image
[0010]
Furthermore, a color dropout processing unit for performing color dropout processing on the input image is provided, and after the form is identified, the color specified in advance is subjected to dropout processing based on the information on the color dropout processing required for the form identification result. Features.
[0011]
DETAILED DESCRIPTION OF THE INVENTION
First Embodiment FIG. 1 is a block diagram showing a first embodiment.
In the figure, reference numeral 1 denotes an image input processing unit which has a function of inputting an image of a form to be input as image data and storing it in the image data storage memory 2. In general, this is a scanner device, FAX, or the like. The image data storage memory 2 is a memory for storing the image data input by the image input processing unit 1.
[0012]
A feature information extraction processing unit 3 has a function of extracting feature information serving as an input for form identification from image data stored in the image data storage memory 2. Generally features ruled line features (information consisting of the shape, length, number, configuration, etc. of ruled lines constituting a form) and area features (information consisting of position information and unique pattern information such as titles on forms) It corresponds to information.
[0013]
Reference numeral 4 denotes a feature information storage memory which stores the feature information extracted by the feature information extraction processing unit 3.
Reference numeral 5 denotes a form format storage memory, which is a form format in which all the form identification feature information to be classified in this system, and information on the position and attributes of areas to be processed such as character recognition of the form are set in advance. A memory for storing information. The form format storage memory 5 stores all or a part of the form format information of the classification target registered in advance as the classification target. Note that only the form identification feature information may be stored. In this case, the form format information relating to the position and attributes of the character recognition target area can be stored in an external storage device such as a disk.
[0014]
Reference numeral 6 denotes a form identification processing unit, which compares the form identification feature information stored in the form format storage memory 5 with the feature information of the individual input images stored in the feature information storage memory 4 to obtain the corresponding form. Has a function to specify or narrow down.
If the form identification is successful, data (form name, ID, etc.) indicating the form of the identification result is stored in the form identification result storage memory 7.
[0015]
An image rotation processing unit 8 refers to the form format storage memory 5 having the form identification feature information from the form identification result stored in the form identification result storage memory 7, obtains information on the image rotation direction, and The image data stored in the data storage memory 2 is rotated in the designated direction.
Reference numeral 9 denotes a post-rotation image data storage memory which stores image data after direction rotation.
[0016]
An area setting processing unit 10 sets an area to be processed on the image data stored in the rotated image data storage memory 9 while referring to the format of the corresponding form stored in the form format storage memory 5. It has a function to do.
An example of a processing target form is shown in FIG. The form 13 is, for example, a form of A4 horizontal size (a direction in which A4 size paper is placed in landscape orientation is a positive direction).
[0017]
The form 14 is a form of A4 portrait size (a direction in which A4 size paper is placed in portrait orientation is the forward direction).
It is assumed that the forms 13 and 14 are read together by a single scanner.
For the purpose of shortening the reading processing time, and for the purpose of facilitating scanner running by preventing problems such as skew and jamming due to mixing of horizontal sizes, the form 14 is set on the scanner in the A4 horizontal direction, and the right or left In the figure, an image rotated 90 degrees to the right is acquired (form 14-B).
[0018]
At the time of format registration, the form 13 extracts and registers feature information (ruled line features, etc.) while maintaining the normal direction. The form 14 acquires the feature information for form identification as an image (form 14-B) rotated 90 degrees to the right (left) in the same direction as when acquiring the image.
The position information of the area to be processed is acquired as a forward image (form 14-A).
[0019]
Hereinafter, the operation will be described with reference to FIG.
S 1, the image input processing unit 1 inputs the image data of the processing target form and stores it in the image data storage memory 2. (Image input)
S 2, the feature information extraction processing unit 3 extracts feature information for identification from the image data of the processing target form stored in the image data storage memory 2 and stores it in the feature information storage memory 4. The feature information includes those that can be extracted from only image data such as ruled line features and those that are extracted based on information registered in the form format such as region information. In the latter case, the form format storage memory 5 is also referred to. (Feature information extraction)
S3, the form identification processing unit 6 sequentially extracts form identification feature information of the processing target form registered in advance from the form format storage memory 5, and collates it with the feature information of the image data stored in the image data storage memory 2 By doing so, it is identified which form the image data is. The identification result is stored in the form identification result storage memory 7. If there is a corresponding form, information (form name, ID, etc.) specifying the form is stored. When there is no corresponding form, information indicating no corresponding such as an error code is stored. (Form identification)
S4, if there is a corresponding form, information indicating whether or not image rotation is necessary is read out from the form format of the form stored in the form format storage memory 5 before the area setting process. That is, if the processing target form is the form 13 shown in FIG. 2, “NO: no image rotation”, and if it is the form 14, “YES: rotate 90 degrees left”.
[0020]
In S5, when image rotation is required, the image rotation processing unit 8 rotates the image in the direction specified by the form format. That is, in the case of the form 14 in FIG. 2, since the image in the direction of the form 14-B is acquired, the image is rotated 90 degrees to the left and returned to the direction of the form 14-A. The rotated image data is stored in the rotated image data storage memory 9.
S6, the area setting processing unit 10 takes out information on the position and attributes of the character recognition target area of the specified form from the form format storage memory 5 and sets the area.
[0021]
S7, character recognition processing is performed on the set area.
According to the embodiment described above, in the form image processing system in which the paper directions of input forms are mixed, image input is performed with the paper directions aligned, and after identifying the form in the same direction, image rotation is required. In the case of a form, the image can be rotated and the subsequent region setting, character recognition, etc. can be performed, so that the occurrence of skew and the like is reduced, and the image acquisition accuracy is improved and the image acquisition time is shortened. Can be measured.
[0022]
In addition, it is possible to read a mixed form of a B4 vertical form such as a B4 horizontal form and a form having a different paper direction.
FIG. 4 shows a block diagram of an apparatus according to the second embodiment. Since the structure is substantially the same as that of the first embodiment, description thereof is omitted.
[0023]
In FIG. 5, a form 15-A is a form in which ruled lines are printed in a blue color, and a form 16-A is a form in which ruled lines are printed in a red color.
It is possible to improve the recognition performance by dropping out the color of the ruled lines printed in these blue and red colors so that the ruled lines cannot be seen.
However, if the ruled line is dropped out, the characteristic of identifying the form such as the form 15-B and the form 16-B is lost. Therefore, at the time of format registration, when extracting feature information for form identification, both the forms 15 and 16 use the images 15-A and 16-A before color dropout processing.
[0024]
The operation will be described below with reference to FIG.
S 1, the image input processing unit 1 inputs the image data of the processing target form and stores it in the image data storage memory 2. At this time, since the color dropout processing is not executed, the images of the forms 15-A and 16-A are input. (Image input)
S 2, the feature information extraction processing unit 3 extracts form identifying feature information from the image data of the processing target form stored in the image data storage memory 2 and stores it in the feature information storage memory 4. The feature information includes those that can be extracted from only image data such as ruled line features and those that are extracted based on information registered in the form format such as region information. In the latter case, the form format storage memory 5 is also referred to. (Feature information extraction)
S3, the form identification processing unit 6 sequentially extracts form identification feature information of the processing target form registered in advance from the form format storage memory 5, and collates it with the feature information of the image data stored in the image data storage memory 2 By doing so, it is identified which form the image data is. The identification result is stored in the form identification result storage memory 7. If there is a corresponding form, information (form name, ID, etc.) specifying the form is stored. When there is no corresponding form, information indicating no corresponding such as an error code is stored. (Form identification)
S4, if there is a corresponding form, information indicating whether color dropout is necessary before area setting, character recognition, etc. is read out from the form format of the form stored in the form format storage memory 5. That is, when the processing target form is the form 15 shown in FIG. 5, “YES: blue dropout”, and when it is the form 16, “YES: red dropout”.
[0025]
S5, when color dropout is necessary, the color dropout processing unit 11 performs color clustering processing of the image with the color specified in the form format. In the case of the form 15, a blue dropout process is performed, and an image such as an image 15-B is stored in the image data storage memory 12 after color dropout. In the case of the form 16, red-type dropout processing is performed, and an image such as an image 16 -B is stored in the image data storage memory 12 after color dropout.
[0026]
S6, the area setting processing unit 10 takes out information on the position and attributes of the character recognition target area of the specified form from the form format storage memory 5 and sets the area.
S7: Character recognition processing is performed on the set area using the image data after color dropout.
[0027]
According to the embodiment described above, in a form image processing system in which dropout colors are mixed in an input form, an image before color dropout processing is input, form identification is performed using the image, and the result of identification is determined. By performing color dropout processing and performing processing such as area setting and character recognition after that processing, the types of processing target forms can be expanded and the accuracy of character recognition can be improved.
[0028]
【The invention's effect】
According to the present invention described in detail above, the image data of all the multiple types of forms to be read are read, and registered in the form format storage memory in advance as form identifying feature information for identifying each form. By comparing the feature information of the image data of the form input to the above and the previously registered form identification feature information, it is possible to identify the individual input forms. In the case of a form with both vertical and horizontal orientations, it is possible to identify individual input images and rotate the image data that needs to be rotated in the forward direction to correct the direction and set the character recognition target area Have
[0029]
In addition, forms that have mixed color display can be processed by color-dropping out the individual input forms, which can increase the types of forms to be processed and improve the accuracy of character recognition. Have
[Brief description of the drawings]
FIG. 1 is a block diagram showing a first embodiment. FIG. 2 is an explanatory diagram of a form. FIG. 3 is a flowchart showing a processing procedure. FIG. 4 is a block diagram showing a second embodiment. [Explanation of form] [Figure 6] Flow chart showing processing procedure [Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Image input process part 2 Image data storage memory 3 Feature information extraction process part 4 Feature information storage memory 5 Form format storage memory 6 Form identification process part 7 Form identification result storage memory 8 Image rotation process part 9 Image data storage memory 10 after rotation Area setting processing unit 11 Color dropout processing unit 12 Image data storage memory 13 after color dropout 13 Form 14 Form 15 Form 16 Form

Claims (2)

Image data of multiple types of forms to be read is read, and ruled line feature information in a state rotated in the forward direction so that the form sizes are aligned as the form identification feature information for identifying each form, together with the rotation direction information in advance. Register in the form format storage memory, identify the form by comparing the feature information of the image data of the individual input form with the pre-registered form identification feature information, and rotate the corresponding form A form reading method characterized in that an image is rotated by information to obtain a forward image, and processing target area information such as character recognition is set for the forward image. Image input means for inputting image data of a form to be read and storing the image data in an image memory;
A form format storage memory in which ruled line feature information in a state rotated from the forward direction so that the form size is aligned as the form identification feature information respectively extracted from the image data of all the handled forms, is registered in advance together with the rotation direction information,
A feature information extraction processing unit for extracting feature information of the input image;
A form identifying means for identifying the form by comparing the extracted feature information with the previously registered form identifying feature information;
It consists of an image rotation processing unit that rotates the input image,
It is characterized by performing form identification, rotating the image according to the rotation direction information of the corresponding form to obtain a forward image, and setting processing target area information such as character recognition for the forward image. Form reading system.

Images