JP4651407B2 - Image processing apparatus, computer program, and storage medium - Google Patents

Description

The present invention holds a designated area designated in a document image as a template, acquires feature information inside and around the designated area for the image for creating the template, holds the feature, and The present invention relates to an image processing apparatus that extracts the same position as the designated area, a computer program, and a storage medium.

  In the existing form processing, the form form is completely fixed, and there are almost fixed characters, ruled lines, frames, timing marks, etc. in any form. There are already many known systems for processing the entry part of such a form.

  For example, a method of performing alignment using cross points where ruled lines intersect (see Patent Document 1), and a method of detecting and correcting image deviation by extracting a predetermined mark in an image by a shape recognition method ( (See Patent Document 2)

  However, as described above, these methods require some fixed mark, ruled line, frame, and character as a reference for alignment. For this reason, it cannot be used when you want to input a lot of things that are written in the same layout, such as the title of a paper, but are not like a form, at high speed.

  As a technique for accurately extracting bibliographic items using such layout information, for example, as disclosed in Patent Document 3, region segmentation is performed on an input image, and features regarding the layout structure are extracted. It is possible to use a method of creating a model and extracting a bibliographic item desired for a model that matches a model created in advance. This method is very effective when retrieving bibliographic items of journals, but it is very useless when there is no need to classify by layout, and it is very dependent on the performance of area identification. However, there is a problem that it is difficult to obtain a stable region identification result, and that it is necessary to create another model if the layout other than the acquisition position is slightly different.

  Further, in the “form data reading method, recording medium, and form data reading apparatus” of Patent Document 4 by the present applicant, the structure recognition of the table and the character recognition result are used to determine what is to the left (or below) of the item. We have proposed a system that can recognize even if the degree of freedom of the form format is increased by describing whether there is an area that should be recognized.

However, for documents that do not have such a high degree of freedom as to be effective with this method, this method for character recognition consumes processing time and resources, and lacks stability because it depends on the OCR result. Furthermore, in a document having a plurality of tables, it is necessary to determine which table should be recognized as a superordinate concept of this method, which cannot be solved only by the proposed method.
Japanese Patent Laid-Open No. 10-091783 JP-A-4-261259 JP 11-328306 A JP 2000-29986 A

The present invention has been made in view of such circumstances, and the order of entry items, the relative position in the table, etc. are limited, but the entered items and structure are relatively the same (quasi-standard). documents, for example, the size of the subtle fonts and frames by the printer differences, slight differences in paper size, resolution scanning difference, or some entry is made to those printed by the printer, scan it After the normal flow of processing, the scanned data is printed again in a later process and added again, and when it is re-scanned, the size slightly changes depending on the margin at the time of scanning and printing. such, it tolerates differences in size, and processing time and required memory capacity images processing device and computer Ru can be suppressed and storage medium providing child The interest.

The present invention holds a designated area designated in a document image as a template, acquires feature information inside and around the designated area for the image for creating the template, holds the feature, and In the image processing apparatus for extracting the same position as the designated area, it is determined whether or not the designated area is a part of a table frame on the template creation image. analyzes the internal tables containing frame surrounding recognizes what position on the structure of the table is specified, so as to retain, determined before SL designated area if it is part of a table frame time, the input request information about the specified area relative to the user is thereby that to refer to the acquired information.

It is also because of the computer program cause the computer to function as the image processing apparatus.

Moreover, it is a storage medium storing the computer program described above .

  Therefore, according to the present invention, alignment of a document that is close to a fixed form is performed, and a specific position of a document that is not completely fixed is used effectively in a system that is created to perform a predetermined process. be able to. For example, it is possible to realize an effect that, for example, it is desired to obtain a part of a table of a document having a certain table structure by alignment.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

  FIG. 1 shows a configuration example of an image processing apparatus according to an embodiment of the present invention.

  In the figure, a CPU (Central Processing Unit) 1 controls the operation of the image processing unit, and a ROM (Read Only Memory) 2 stores programs executed by the CPU 1 at startup and necessary data. A RAM (Random Access Memory) 3 is for configuring a work area of the CPU 1 and the like.

  The character generator 4 is for generating graphic character display data, the clock circuit 5 is for outputting the current date and time information, and the local area network interface circuit 6 is for localizing this image processing device. The local area network transmission control unit 7 is used for connection to an area network (not shown), and the local area network transmission control unit 7 performs various types of data for exchanging various data with other data terminal devices via the local area network. This is for executing the communication control process.

  The magnetic disk device 8 is for storing various data such as various application programs, work data, file data, image information data, and the CD-ROM device 9 is a CD-ROM which is an exchangeable recording medium. The CRT screen display device 11 is used to read data in the ROM 10, and the CRT screen display device 11 is used to display a screen for operating the image processing device. The display control unit 12 is a display of the CRT screen display device 11. It is for controlling the contents.

  The keyboard device 13 is for performing various key operations on the image processing device, and the screen instruction device 14 is for performing operation operations such as instructing an arbitrary point on the CRT screen display device 11. Yes, the input control unit 15 is for capturing input information of the keyboard device 13 and the screen instruction device 14.

  The scanner 16 is for reading and inputting a document image at a predetermined resolution.

  These CPU 1, ROM 2, RAM 3, character generator 4, clock circuit 5, local area net transmission control unit 7, magnetic disk device 8, CD-ROM device 9, display control unit 12, input control unit 15, and scanner 16 Are connected to the internal bus 17, and data exchange between these elements is mainly performed via the internal bus 17.

  In the present invention, for example, for image data obtained by reading a form document, image processing for extracting specific items of a table, for example, description contents of an issue date, description contents of a title, and the like is performed. At that time, for the extraction item, for example, a document image (template creation image) read by the user is displayed, and an area including a desired extraction item is designated by a user operation. Then, template information representing the designated area is created, and extracted items are extracted based on the created template information.

  FIG. 2 shows a main part of an example of template creation processing according to an embodiment of the present invention.

  First, a template creation image is input (process 101), the input image is displayed (process 102), and the user is allowed to select a designated area (process 103). Also, an inquiry is made as to whether or not the designated area designated by the user is surrounded by a table.

  If the result of determination 104 is YES in the case of being surrounded by a table, the structure inside the table including the designated area is analyzed (processing 105). Next, the position of the specified area is recognized in the table structure (process 106), and the template dictionary (not shown) for storing the template stores the “specified in the table” and “table structure” for the specified area. "In which position" and "table structure analysis result" are registered (process 107), and this process ends.

  If the result of determination 104 is NO, it is registered in a normal template dictionary (process 108), and this process ends.

  FIG. 3 shows a main part of an example of a template creation process according to another embodiment of the present invention. In this case, the specified area is determined after the table is extracted in advance for the entire image.

  First, a template creation image is input (process 201), the input image is displayed (process 202), and the user is allowed to select a designated area (process 203).

  Next, a table included in the entire input image is extracted (process 204), and the structures inside all the found tables are analyzed (process 205).

  Then, it is checked whether or not the designated area is surrounded by a table (decision 206). At this time, the user can be inquired as to whether or not the designated area is surrounded by a table.

  If the designated area is surrounded by a table and the result of determination 206 is YES, the template dictionary (not shown) for storing the template stores “there is in the table”, “table structure” for the designated area. "In which position" and "table structure analysis result" are registered (process 207), and this process ends.

  If the result of determination 206 is NO, it is registered in a normal template dictionary (process 208), and this process ends.

  In this way, in this case, since the structural analysis is performed on all the tables included in the image first, the processing efficiency increases when a plurality of areas are set.

  FIG. 4 shows the main part of an example of template creation processing according to still another embodiment of the present invention.

  First, a template creation image is input (process 301), the input image is displayed (process 302), and the user is allowed to select a designated area (process 303).

  Next, a table included in the entire input image is extracted (processing 304), and it is checked whether or not the designated area is surrounded by the table (decision 305). At this time, the user can be inquired as to whether or not the designated area is surrounded by a table.

  If the result of determination 305 is YES in the case of being surrounded by a table, the structure inside the table in which the specified area is surrounded is analyzed (processing 306). Next, register “being in the table”, “in which position in the table structure”, and “table structure analysis result” for the designated area in the template dictionary (not shown) for storing the template. (Process 307), this process is terminated.

  When the result of determination 305 is NO, it is registered in a normal template dictionary (process 308), and this process ends.

  Here, the determination of whether the designated area is surrounded by the table can be automatically processed based on the input image. For example, assuming that the size of the designated area is width W and height H, it is only necessary to search whether there is a black run having a ratio equal to or greater than a certain threshold of W, for example, a length of 100% or more above and below the designated area. . Since the designated area is usually set larger than the part where characters are written (especially when a form on which characters are already written is used as a template), such a method may be used. Note that if the threshold value is too small, characters such as “1” may be reacted in the vertical direction and underlined in the horizontal direction, so the setting needs to be taken care of.

  The position of the table structure is the table coordinate value on the template, the position of the center of gravity of the table, the coordinate value of the specified area, the position of the center of gravity of the specified area, the position of the ruled line surrounding the specified area, and its center of gravity. , The position of the designated area and the position of the center of gravity in the relative coordinate values when the starting point of each table is considered as the origin, and information on the cell ID of the table.

  Regarding the cell ID, refer to Japanese Patent Laid-Open No. 2003-223611. By using this cell ID, the degree of freedom of the relative position of the coordinate increases, and if the structure (number of ruled lines, number of cells) is the same, the cell ID value is the same even if the frame size changes. Since it is a value, the designated area can be appropriately determined.

  FIG. 5 shows an example of processing for determining the designated area included in the input image using the template dictionary created in this way.

  First, a template dictionary is read (process 501), an image to be recognized is input (process 502), and an included table is extracted for the entire input image (process 503). Then, the structure of all the extracted tables is analyzed (process 505), collated with the table structure position of the template (process 505), and an area designated by the user is extracted (process 506).

  Then, the designated process is applied to the extracted area (process 507).

  Here, in order to determine whether or not there is a table in the input image to be recognized, for example, as described in Japanese Patent No. 2940936, a circumscribed rectangle is extracted, and a run with a certain length or more inside is extracted. There is a known method such as a method of extracting a rectangle formed by the above as a ruled line and determining a table based on the number of extracted ruled lines, which is also available here.

  In the table structure analysis method, for example, as described in JP-A-4-167194, ruled line extraction in the main scanning direction and ruled line extraction in the sub-scanning direction are performed, and information on ruled lines and frames is analyzed. There is a way. Alternatively, even a table without ruled lines can be analyzed by a method of virtually creating ruled lines as in the method described in Japanese Patent No. 28129982, and dotted lines are also described in Japanese Patent No. 3140079. By applying the method described above, it is possible to perform extraction using a histogram of the lengths of black runs and white runs. The table determination function and the table processing function are configured by combining such existing methods. It is also possible to use known methods not described here.

  In the table structure collation method, the table structure position stored in the template dictionary is compared with the table structure extracted from the input image. This includes, for example, the coordinate value of the table on the template, the position of the center of gravity of the table, the coordinate value of the specified area, the position of the center of gravity of the specified area, the position of the ruled line surrounding the specified area, and the position of the center of gravity. At least one of the position of the designated area and the position of the center of gravity in the relative coordinate values when the starting point of each table is considered as the origin, and information on the cell ID of the table can be adopted.

  In addition, information that is surrounded by a table frame is important, and information that there is no table frame in the vicinity is also very important information for realignment. For example, if there is only one frame in the search range, even if the position is slightly shifted, the possibility that the position of the frame matches the desired position is very high.

  On the other hand, when there are many table frames in the vicinity, a slight shift can cause the adjacent frame and the desired frame to be mistaken.

  Therefore, even if the specified area is inside the table frame, if there is no other table frame in the vicinity, control to increase the search margin, and conversely, the specified area is the table frame and there are several table frames in the vicinity. In this case, it is possible to perform collation of the table structure with higher accuracy by controlling the search margin to be small.

1 is a block diagram showing a configuration example of an image processing apparatus according to an embodiment of the present invention. The flowchart which showed the principal part of an example of the template preparation process concerning one Example of this invention. The flowchart which showed the principal part of an example of the template preparation process concerning the other Example of this invention. The flowchart which showed the principal part of an example of the template preparation process concerning the further another Example of this invention. The flowchart which showed an example of the process at the time of determining the designated area | region contained in an input image using a template dictionary.

Explanation of symbols

1 CPU (Central Processing Unit)
2 ROM (Read Only Memory)
3 RAM (Random Access Memory)
11 CRT screen display device 13 Keyboard device 14 Screen pointing device 16 Scanner

Claims (3)

The designated area designated in the document image is held as a template, and the feature information of the inside and the surrounding of the designated area is acquired for the image for creating the template, the feature is held, and the designated area for the input image In the image processing apparatus that extracts the same position as
On the template creation image, it is determined whether or not the designated area is a part of a table frame. If the designated area is a table, the inside of the table including the frame surrounding the designated area is analyzed to determine the structure of the table. Recognize and hold which position is specified above,
When the designated area is to determine whether a part of the table frame, enter requests information about a specified area relative to the user, the image processing apparatus characterized by referring to the information acquired by it. Because the computer programs cause the computer to function as the image processing apparatus according to claim 1. A storage medium storing the computer program according to claim 2 .

Images