JPH1063869A - Document processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce user's correcting operation in total by showing a correction place by similar area extraction and making a user confirm whether or not a correcting process is correct. SOLUTION: A similar area extraction part 205 extracts a similar area having similar structure from a registered document stored in a structure storage part according to frame information extracted by a structure analysis part 204. A structure substitution part 206 extracts differences between the structure having the similar area extracted by the extraction part 205 and the part corresponding to the similar area extracted by the analysis part 204 by making them correspond to each other and corrects the parts where the differences are present by substitution processing, etc. Further, a structure correction part 207 shows a corrected document obtained by the substitution part 206, receives user's confirmation input or indication input for manual structure correction by the user, and makes correction according to the indication input. Thus, the correcting process can be performed by the user's simple procedure.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a document processing apparatus for capturing images of various documents and converting the images into data.

[0002]

2. Description of the Related Art Some conventional document processing apparatuses recognize a document structure by inputting a document image and performing image processing to estimate ruled lines and character areas. In this type of document processing device,
First, a document in a paper state is input as image data using an image input device such as a scanner. The input image data is subjected to image processing such as binarization, thinning, and line segment approximation, and is extracted as a line segment. Then, it classifies into ruled lines, character regions, and other components by using feature amounts such as the length, directionality, and distance between the extracted line segments.

[0003] Conventional document processing devices include:
When a ruled line is combined and extracted as a frame, and there is a character area within or near the frame, the character area is converted to a character code by performing character recognition, and the meaning of the frame is determined. There is something that understands the structure.

[0004] Conventional document processing devices include:
In some cases, a plurality of character regions are integrated in consideration of the size of the character regions and their distance from each other, and are extracted as a series of character regions. Further, there is a method in which a paragraph region is defined in consideration of a mutual positional relationship between the regions, and a layout structure of the entire document is obtained as document structure information.

[0005]

However, in the conventional document processing apparatus as described above, erroneous recognition or erroneous determination exists in most of the processing, and the erroneously set portion is included in the obtained document structure information. Many are seen. This is because a noise component is added to the input document image or a plurality of structural interpretations exist only in the physical state of the document, and it is impossible to determine an ambiguous part. Therefore, it requires a considerable amount of correction work by the user.

Japanese Unexamined Patent Publication No. Hei 7-220023 discloses a technique for accurately obtaining format information corresponding to a tilt at the time of input. However, this technique only corrects misregistration of a ruled line. Therefore, it is impossible to correct the detailed setting regarding the entry area of the document including the entry portion represented by the form or the like.

[0007] The present invention has been made in consideration of the above circumstances, and can automatically correct a misrecognized portion in recognizing document structure information, thereby greatly reducing the amount of correction work by a user. It is an object to provide a document processing device.

[0008]

SUMMARY OF THE INVENTION According to the present invention, there is provided a document processing apparatus for performing predetermined processing by inputting an image of a document, comprising: structure storage means for storing the structure information of the document; A similar area detecting means for searching a document stored in the structure storing means for an area similar to the set structure information of the input document; and a search for the structure information of the input document. And a structure correcting means for correcting the structure information of the input document based on the difference from the obtained similar region structure information.

Preferably, the structure correcting means sets, as a corrected portion, a portion of the input document structure information in which a difference exists between the structure information of the input document and the structure information of the searched similar area. Correction location setting means for performing the correction by replacing the structural information corresponding to the correction location set by the correction location setting means with the structural information of the similar area. I do.

[0010] Preferably, the structure correcting means includes a correction point presenting means for presenting the correction point set by the correction point setting means, and a correction confirming means for accepting a confirmation input as to whether or not the correction is correct for the presented correction point. Wherein the structure replacement unit performs the correction when receiving a confirmation input from the correction confirmation unit that the structure should be corrected.

Preferably, the structure modifying means presents modified structure creating means for creating structure information modified using the structure information of the similar area as a modification plan, and presents the structure information as the created modification plan. And a correction confirmation means for accepting a confirmation input on whether or not modification of the displayed structural information is correct, and receiving a confirmation input to the effect that modification is required from the modification confirmation means. It is characterized in that a correction is made in the event of a failure.

Preferably, the apparatus further comprises block extracting means for dividing the document into blocks, wherein the similar area detecting means searches for similar areas in block units.

Preferably, data entry area extraction means for extracting a data entry area to be input by handwriting from the input document image, character recognition means for extracting a character string from the input document image, The data entry area further includes a related character string extraction unit that associates a character string that describes the data entry area among the extracted character strings, and the block extraction unit includes the associated character string as the The block is extracted by integrating data entry areas.

The invention relating to each of the above-described devices is also valid as an invention relating to a method. ADVANTAGE OF THE INVENTION According to this invention, the incorrect part in recognition can be easily corrected, and the amount of correction work by a user can be reduced.

Further, according to the present invention, it is possible to make a correct correction by a simple procedure of the user by asking the user whether the correction of the correction portion presented is correct or not, and to improve the user's entirety. Correction work can be reduced.

[0016]

Embodiments of the present invention will be described below with reference to the drawings. First, a document processing apparatus according to a first embodiment of the present invention will be described.

In this embodiment, the format (for example, a blank form in which data is not entered in the data frame) of each document is stored in a database, and the document (for example, a data form in which data is entered in the data frame) is read. In this case, a document processing apparatus that performs a processing procedure registered in advance according to the format, in which the present invention is applied to a part (a structure registration unit 102 described later) that performs a format registration process, will be described as an example. I do.

FIG. 1 is a diagram showing a configuration of a document processing apparatus according to the present embodiment. This document device
Image input unit 101, structure registration unit 102, structure storage unit 10
3, a procedure setting unit 104, a structure identification unit 105, a procedure processing unit 106, and a data output unit 107.

The image input unit 101 is for taking in an image of a document as digital data, and is constituted by using, for example, a scanner. The structure registration unit 102 recognizes the format of the input document (blank form) using image processing and graphic recognition, and recognizes a part that cannot be recognized (a part whose recognition result is different from the registered structure). The structure information is set by performing the correction in ()).

The structure storage unit 103 stores structure information of a document format as a database. The procedure setting unit 104 sets what processing should be performed when a document (data form) of the same type as a document (blank form) stored as a database is input. This setting is performed, for example, for each registered document or for a certain document group.

When a document (data form) to be processed is input in the image input unit 101, the structure identification unit 105 registers a document of the same type as the input document in a database in the structure storage unit 103. From what you are doing.

The procedure processing unit 106 performs the procedure set by the procedure setting unit 104 according to the type of the identified document. The data output unit 107 outputs the data processed by the procedure processing unit 106 to a screen or a file.

FIG. 2 shows a structure registration unit 10 according to this embodiment.
2 is a diagram illustrating an example of the internal configuration of FIG. Main structure registration unit 10
2 is an image processing unit 201, a graphic processing unit 202, a character recognition unit 203, a structure analysis unit 204, a similar region extraction unit 205,
A structure replacement unit 206 and a structure correction unit 207 are provided.

The image processing unit 201 extracts a line segment and a character area from an image of a document (blank form) input by the image input unit 101. Figure processing unit 2
02 classifies the extracted line segment, extracts a frame, and extracts a character area.

The character recognition unit 203 converts the character image in the extracted character area into a character code. Structural analysis unit 20
4 acquires the detailed information of the extracted frame.

The similar area extracting unit 205
4 based on the frame information extracted by
A similar area having a similar structure is extracted from the registered documents stored in the storage area 03.

The structure replacement unit 206 includes the similar area extraction unit 20
5 and a portion corresponding to the similar region in the structure extracted by the structure analysis unit 204, thereby extracting a difference between the structures. The existing part is corrected by a replacement process or the like.

The structure correcting unit 207 presents the document after the correction by the structure replacing unit 206, receives a user's confirmation input (when it is correctly recognized), or inputs an instruction for manual structure correction by the user. (add to,
Deletion, replacement, etc.) are accepted, and correction is performed based on the instruction input (when an erroneous portion exists).

Hereinafter, the format registration processing will be described with reference to FIG. FIG. 3 is a flowchart illustrating an example of a format registration process of each document performed by the structure registration unit 102 according to the present embodiment.

First, when a document image is input for format registration (step S210), a line segment is extracted in the image processing unit 201 (step S21).
1). Here, line elements are extracted by performing thinning processing or contour tracing processing on the input binary image, and all image components are converted into line segments by performing line segment approximation processing.

Next, the graphic processing unit 202 extracts a rectangular frame from the extracted line segment (step S21).
2). As an example of extracting a rectangular frame, Japanese Patent Application No. 7-28
333. In this method, a rectangular frame is extracted by measuring intersections from line segments that intersect perpendicularly to each other, and examining whether a rectangle is formed by these four points when four points are extracted from these intersections. is there. At the same time, ruled lines and underlines are extracted from the remaining line segments not used for the rectangular frame by using their length and directionality.

A short line segment not used for a frame or a ruled line is determined as a line segment forming a character, and a character region is extracted from the distribution state of the remaining line segments (step S21).
3). Then, the character recognition unit 203 converts the image of the printed characters present in the extracted character area into a character code by using a pattern recognition technique (step S214).

Next, in the structural analysis unit 204, the extracted frame is determined by using surrounding frames and character information to determine whether the frame includes an entry portion, and what kind of entry portion is included in the frame. The structure analysis process of this document is performed by extracting attributes such as the type of character to be entered or the meaning of words to be entered (step S215). As the structure analysis processing, for example, the one disclosed in Japanese Patent Application No. 7-28334 can be used.

When the structural analysis process is completed, the process of recognizing the format information of the input document has been completed. However, it is common that there is an error in the recognition result in the input document. This is because there are places where detailed determination of the structural analysis cannot be made only from the information of the image, or the original image is missing due to the influence of noise or the like.

In this embodiment, in order to cope with this,
The format information of the registered document is used as a dictionary at the time of format setting. In order to do this, first, a similar area of the document whose input is currently set is extracted from the registered documents registered in the structure storage unit 103 (step S216).

Here, a method will be described in which, instead of acquiring similar information in page units of the document, the document is divided into regions depending on the contents and similar information is extracted in units of regions.

In order to divide the document into regions depending on the contents, it is assumed that a block is defined in the document, which is extracted and registered at the time of document registration. An example of a process of extracting a block from a document and registering the block as a database will be described later with reference to FIG. An example of a process for extracting a similar region in block units will be described later with reference to FIG.

When a similar region of a document is extracted, a difference from the extracted similar region is detected, and information of the similar region is replaced with respect to a portion where the difference occurs, thereby performing automatic correction ( Step S217). An example of the correction process performed here will be described later with reference to FIG.

If there is any part that cannot be automatically corrected, the user can manually correct it.
The corrected format information is stored in the database as a new format (step S21).
8).

Thus, the document format registration processing is completed. Next, a process of extracting a block from a document and registering it as a database will be described with reference to FIG. FIG. 4 is a flowchart illustrating an example of a process for extracting a block using the set document format information and registering the extracted block as a part of a database.

This processing is performed at the stage when all the format settings have been completed for the input document in the structure storage unit 103 of FIG. First, information on the format for which setting has been completed is input (step S30).
1).

The information used for extracting a block is information on a character string associated with each frame extracted in the format setting. Small blocks are extracted by combining frames that associate the same character string (step S302). At this time, the extracted block is set to a level 0 block except for a special case described below.

Next, by combining a plurality of the extracted small blocks, those forming a quadrilateral are extracted as larger blocks (step S303).
Each time you combine blocks, raise the level of the block,
Make each have a parent-child relationship. by this,
You can create a hierarchical structure for blocks. With respect to all the extracted large and small blocks, the coordinates of the area occupied by the blocks, information on the frames existing in the blocks, and the character strings are measured (step S304).

Finally, the information of these blocks is added to the link of the block level and the parent-child relationship and registered in the database (step S305). At this time, the existence of the same block is checked based on the relative position of the frame and the content of the character string so that the block having the same structure is not registered twice.

Here, a method of extracting a small block will be described using a specific example with reference to FIG. The small block is
As described above, the same character string is extracted by combining the associated frames.

For example, as shown in FIG.
Is associated with the character string 503, and the character string 503 is associated with the frame 502. At this time, the range indicated by the area 504 is extracted as a block. The level of this block is 0.

Further, as shown in FIG.
Character strings 508 and 510 for the frame, character strings 509 and 510 for the frame 506, and a character string 50 for the frame 507.
9 and 510 are associated with each other.
The region 512 and the region 5
Considering that there are two blocks 13, the area 511 is extracted as a level 1 block and the areas 512 and 513 are extracted as level 0 blocks, and a parent-child relationship is defined between them.

Next, referring to FIG.
An example of the similar region extraction processing in No. 16 will be described. First, structural information relating to the extracted frame and character string is input (step S401), and a minimum block is extracted (step S402). As a method of extracting the minimum block, the method of extracting the small block of level 0 described with reference to FIG. 5 is taken. Blocks are extracted by integrating the frames using the information on the character strings associated with the frames, and the extracted blocks are defined as the minimum blocks.

Block search processing is repeatedly performed so that similar regions are extracted from all of the extracted minimum blocks (step S403). Extraction of similar regions
The process starts with a level 0 block search process for the extracted minimum block. The same block is detected from the level 0 blocks registered in the database, and is extracted as a similar block (step S404). The determination of the same block is made based on the connection state of the frame in the block and the content of the character string.

If a similar block has not been extracted, the process from step S403 is performed on the next smallest block (step S405). If a similar block is extracted, the information of the block one level higher is extracted using the parent-child relationship of the block (step S40).
6).

If the upper block has not been detected (NO in step S407), the extraction processing of the similar area is terminated, and the area found so far is set as a similar area (step S410).

If an upper block is detected (YES in step S407), the frame and character string existing in the detected upper block are associated with the input document frame and character string, respectively. The similarity is calculated from the number of successful attachments (step S408). The similarity is defined by, for example, (number of successful associations) / (total number of frames and character strings). If the similarity is larger than a certain threshold value α (for example, 0.7), it is determined that a similar region has been found, and the process returns to step S406 and is repeatedly performed to extract a similar region larger than this. Step S40
9).

If the similarity does not reach the threshold value α, the area found so far is set as a similar area (step S410). If no similar area is found,
It is assumed that a similar area could not be extracted for the current target minimum block. That is, it means that a document having a similar format has not been registered.

In the above description, even when a plurality of similar regions may exist, the first extracted one is determined as the similar region. Instead, the similar region is determined from the structural information registered in the database. All of them may be extracted, and the one having the highest number of levels and the highest similarity may be determined as a similar area.

Next, referring to FIG. 7, step S21 will be described.
7 will be described. First, an association list is created (step S412) from information on primitives (frames, character strings, character codes, entry areas, etc.) in the similar area corresponding to the correction target range (step S411).

According to this association list, each primitive in the correction target range is associated with a primitive in the similar area (step S413). A correction method such as replacement, deletion, or addition is selected according to the result of the association and the type of the primitive, with the primitives that could not be associated and the primitives that were not in one-to-one correspondence selected as correction targets (step S414). ).

Finally, the data is corrected according to the selected correction method (step S415). For example, among the primitives that could not be associated, the primitive data that exists only in the correction target is deleted, and the primitive data that exists only in the similar area is added to the primitive data. In addition, primitives that correspond in many-to-one correspondence make corrections to integrate data, and primitives that correspond in one-to-many make corrections to divide data. In addition, for the size, measures such as equal division are applied.

For example, among primitives that are associated one-to-one, if a difference is found in the character code, the change is made by replacing the character code. At this time, by using the similarity at the time of character recognition, it is possible to limit the correction so that the correction is performed only on the one with the low similarity.

As described above, the similar area is searched from the database using the format recognition result, and the recognition result is compared with the searched similar area, whereby the automatic correction is performed according to the already set format information. It can be carried out.

Next, a document processing apparatus according to a second embodiment of the present invention will be described. This embodiment presents a correction plan to the user before performing the automatic correction,
The automatic correction is made after receiving the confirmation input.

The configuration of the document processing apparatus of this embodiment is the same as that of the first embodiment except for the part relating to the "structure registration unit", and therefore, detailed description thereof is omitted, and the first embodiment is described here. The part related to the “structure registration unit”, which is different from the above, will be mainly described.

The configuration of the document processing apparatus of this embodiment is basically the same as that of FIG. 1, but there is a portion added to the internal configuration of the structure registration unit. The same applies to each processing other than the structure modification processing described in the first embodiment. For example, the flowcharts of FIGS. 3, 4, and 6 can be used.

FIG. 8 shows an example of the internal configuration of the structure registration unit 102 of the document processing apparatus according to this embodiment. The structure registration unit 102 includes an image processing unit 201, a graphic processing unit 20
2, a character recognition unit 203, a structure analysis unit 204, a similar area extraction unit 205, a structure replacement unit 206, a structure correction unit 207, a structure display unit 208, and a correction confirmation unit 209. That is, the configuration is such that the structure display unit 208 and the correction confirmation unit 209 are added to the structure registration unit in the first embodiment.

The image processing unit 201 extracts line segments from a document image input by the image input unit 101,
Extract the character area. The graphic processing unit 202 classifies the extracted line segment, extracts a frame, and extracts a character area.

The character recognition section 203 converts the character image in the extracted character area into a character code. Structural analysis unit 20
4 acquires the detailed information of the extracted frame.

The similar area extracting unit 205 is configured to
4, the structure storage unit 10 based on the frame information extracted by
3. A similar area is extracted from the registered document stored in No.3.

The structure replacing section 206 is similar to the similar area extracting section 20.
The difference is extracted by associating the structure of the similar region extracted by step 5 with the structure of the corresponding region extracted by the structure analysis unit 204, and the difference is extracted. After that, the part where the difference exists is corrected by a replacement process or the like.

The structure correction unit 207 receives a manual instruction input (addition, deletion, replacement, etc.) for the structure correction by the user, and performs correction based on the instruction input.
The structure display unit 208 displays the set structure, the structure of the automatically corrected symmetric portion, the automatically corrected structure, or the automatic correction plan.

The correction confirming unit 209 accepts an instruction input relating to automatic correction, such as an input of confirmation of automatic correction or cancellation by the user. Next, the structure correcting process in step S217 will be described with reference to FIG.

FIG. 9 is a flowchart showing an example of the structure correcting process in the document processing apparatus according to the present embodiment. A correspondence list is created (step S412) from information on primitives (frames, character strings, character codes, entry areas, etc.) in the similar area corresponding to the correction target range (step S411).

According to this association list, each primitive in the correction target range is associated with a primitive in the similar area (step S413). As described above, for the primitives that could not be mapped and the primitives for which the mapping was not one-to-one, correction methods such as replacement, deletion, and addition were selected according to the mapping result and the type of primitive. (Step S414).

Next, in the present embodiment, an object modified according to this modification method is generated in the work area and displayed on the structure display unit 208 (step S416). After confirming whether the modification format displayed to the user is valid or not in the modification confirmation unit 209 (step S417), if confirmation [OK] for modification is obtained (step S418), the actual Correction processing is performed on the data of.

If the correction has been canceled (step S418), the same processing is continued for the next similar area without performing the correction. A specific example of the correction processing is as described with reference to FIG.

In the above procedure, even when a plurality of similar regions may exist, the first extracted one is presented as a similar region, and a correction plan is presented. It is also possible to extract all similar regions from the information, create a correction plan with the highest number of levels and the highest similarity as a similar region, and present it.

When there are a plurality of similar regions, some correction plans may be presented. in this case,
Some of the similar areas having the highest number of levels and the highest similarity may be selected in order from the similar area, and correction plans may be respectively created, and these may be sequentially presented.

Hereinafter, some specific examples of the above processing will be described. FIG. 10 is an example of a format correction example screen displayed on the structure display unit 208. Note that 60 in FIG.
0 indicates the entire screen, and 601 indicates a display window displayed on the screen. In this example, when a difference from the similar region is found in the recognized structure information, a display window 601 is generated, and the corrected result in which the recognized structure information is subjected to a replacement process or the like on the recognition result 602 is obtained by modifying the correction result. 60
3 are displayed, and the user can select whether to correct this portion by using buttons 604 and 605. By clicking the button 604 when the correction is OK and clicking the button 605 when the correction is NG, the correction processing to the actual data can be controlled.

FIG. 11 is an example of a format correction confirmation screen (display window) displayed on the structure display unit 208 when a plurality of format correction plans are prepared. By preparing a plurality of format amendments, the user can select and amend the plurality of amendments. For example, with respect to the format information 611 of the similar area extracted by the similar area extracting unit 205 as shown in FIG. 11B, a difference from the recognition result 602 is extracted by the structure correcting unit 206, and a plurality of differences are extracted. In such a case, a plurality of correction plans can be created depending on whether or not each is to be corrected.

First, as shown in FIG. 11A, a first modification 613 is displayed in the display window 612. To display another correction plan, the user clicks the button 614 or the button 615. A button 614 displays a previous candidate for a previously displayed revision. Button 615
Displays the next candidate of the amendment proposal. If the currently proposed amendment is OK, the user clicks an OK button 604 to perform an amendment process to the actual data using the amendment currently displayed. If all the corrections are NG, the user can click the button 605 to cancel the correction at that point. When the previous candidate button 614 and the next candidate button 615 are pressed, the display of the modification plan 613 is switched to another modification plan 617 in the window 612.

In this example, as a difference between the recognition result 602 and the extracted similar area 611, three places of a character code “order slip”, a character code “coat”, and a frame (to the right of the coat) are extracted. An amendment can be created by combining the differences.

FIG. 12 shows an example of a format correction confirmation screen (display window) displayed on the structure display section 208 when a format correction portion is displayed. The screen 612 of (a) shows a state in which the recognized format structure is displayed, and when the correction portion is extracted, the display of the primitive at the position to be corrected as shown in (b) is highlighted. As a method of emphasis, for example, color display or blink display is performed.

It is also possible to create an image of a correction plan and display the recognition result and the image of the correction plan alternately. In this case, the recognition result is always displayed in black and the correction plan is displayed in red, so that it is easy to understand which image is currently displayed.

When there are a plurality of portions to be corrected, a portion to be corrected may be selected and instructed by, for example, a user clicking a corresponding portion on the screen with a mouse. In the example of FIG. 12, the character code "order slip" is not selected, and the character code "coat" and a frame (right of the coat) are selected.
Is selected, a correction result as shown in (c) is obtained.

As described above, the similar area is searched from the database using the format recognition result, and the recognition result is compared with the searched similar area to determine whether the correction method is correct. By checking what is displayed, a smooth correction operation can be performed. The present invention is not limited to the above-described embodiment, and can be implemented with various modifications within the technical scope.

[0084]

According to the present invention, when registering the format information of a document, the recognition error in the format recognition processing is automatically corrected by using the format information of the registered document as a dictionary. It is possible to reduce the amount of work required for the user to correct a wrong part.

Further, by presenting the corrected portion by the similar region extraction and allowing the user to confirm whether or not the corrected process is correct, the corrected process can be accurately performed by a simple procedure of the user, and the entire user can be corrected. Correction work can be reduced.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of a document management apparatus according to an embodiment of the present invention.

FIG. 2 is a view showing an example of a configuration of a structure registration unit of the document management apparatus according to the embodiment;

FIG. 3 is a flowchart showing a procedure of a format registration process;

FIG. 4 is a flowchart showing a procedure of a block registration process;

FIG. 5 is a diagram for explaining a small block extraction method;

FIG. 6 is a flowchart illustrating a procedure of a similar region extraction process.

FIG. 7 is a flowchart showing the procedure of a structure modification process;

FIG. 8 is a diagram showing another example of the configuration of the structure registration unit of the document management device according to the embodiment;

FIG. 9 is a flowchart showing a procedure of a structure correction process adopting a correction confirmation;

FIG. 10 is a diagram showing an example of a display screen for confirming format correction.

FIG. 11 is a diagram showing an example of a display screen for confirming format correction.

FIG. 12 is a diagram showing an example of a display screen of a format correction location.

[Explanation of symbols]

 Reference Signs List 101 image input unit 102 structure registration unit 103 structure storage unit 104 procedure setting unit 105 structure identification unit 106 procedure processing unit 107 data output unit 201 image processing unit 202 graphic processing unit 203 character recognition unit 204: structure analysis unit 205: similar area extraction unit 206: structure replacement unit 207: structure correction unit 208: structure display unit 209: correction confirmation unit

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Toshiaki Ishii 1-3-1 Uchisaiwaicho, Chiyoda-ku, Tokyo Tokyo Electric Power Company (72) Inventor Takuo Yamamoto 1-3-1 Uchisaiwaicho, Chiyoda-ku, Tokyo Tokyo Electric Power Company

Claims (6)

[Claims] 1. A document processing apparatus for performing predetermined processing by inputting an image of a document, comprising: a structure storage unit configured to store structure information of the document; and a structure information setting unit configured to set the structure information of the input document input by the image. A similar area detecting means for searching an area similar to the set structure information of the input document from a document stored in the structure storage means; and a structure information of the input document and a structure information of the searched similar area. And a structure correcting means for correcting the structure information of the input document based on the difference between the two. 2. The structure correcting means sets, as a corrected part, a part of the structure information of the input document where a difference exists between the structure information of the input document and the structure information of the searched similar area. A modification portion setting unit; and a structure replacement unit that performs modification by replacing the structure information corresponding to the modification portion set by the modification region setting unit with the structure information of the similar region. The document processing device according to claim 1. 3. The structure correcting means includes: a correcting part presenting means for presenting the correcting part set by the correcting part setting means; 3. The document processing apparatus according to claim 2, further comprising: when receiving a confirmation input from the modification confirming unit that the modification is to be performed, the structure replacing unit performs modification. 4. The modified structure creating means for creating modified structure information as a modified plan using the structural information of the similar area as a modified plan, and presenting the structured information as the modified modified plan. Correction structure display means, and correction confirmation means for accepting a confirmation input of the correctness of the modification with respect to the displayed structural information as a modification plan, and receiving a confirmation input to the effect that the modification confirmation means requires modification. 2. The document processing apparatus according to claim 1, wherein correction is performed in the case. 5. The document processing apparatus according to claim 1, further comprising block extracting means for dividing the document into blocks, and wherein the similar area detecting means searches for similar areas in block units. 6. A data entry area extracting means for extracting a data entry area to be handwritten from an input document image, a character recognizing means for extracting a character string from an input document image, and A data entry area, and associated character string extraction means for associating a character string describing the data entry area among the extracted character strings, the block extraction means further comprising: The document processing apparatus according to claim 5, wherein the block is extracted by integrating entry areas.