JP2009145963A - Document processor and document processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a document processor and a document processing method, allowing high-accuracy extraction of a content area of a heading, a title or the like. <P>SOLUTION: The content area is extracted in line units (step S106). A browsing path is set inside a document image (step S108). A position with the browsing path of each extracted content area as a reference and a position on a page are acquired (step S110). An attribute of each extracted content area is obtained (step S112). The whole of the content area is classified into at least one group by use of attribute information of each content area (step S114). A statistic of the content area in the whole document is calculated (step S116). A statistic of the content area about each group is calculated (step S118). The kind of each group is decided (step S120). Bookmark data showing a position inside the document image of each content area belonging to the group are generated according to the decided group (step S112). <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a document processing apparatus and a document processing method for processing an electronic document including a document image, and more particularly to a technique for generating browsing navigation information for a content area included in a document image.

  As for resource saving, from the viewpoint of space saving, a document management system that converts and manages a document written on a paper manuscript or the like into an electronic document has been put into practical use. In such a document management system, a document image is generated by reading a document using a scanner or the like, and an electronic document is generated from these document images.

  Since such a document image stores the document as an image (image) that is a set of pixels, data (typically, text data, etc.) for specifying the contents such as character strings and charts included in the original document Does not contain. In addition, a document described on a paper manuscript or the like is not necessarily designed so that it can be used electronically. Therefore, when browsing digitized text using browsing software (viewer) on a computer that cannot express all document images at once, the user needs to search (switch the expression screen) in an exploratory manner. It was a very time-consuming task.

  Application software that allows users to interactively set browsing navigation information (typically "bookmarks" or "electronic bookmarks") to places of note on document images for such electronic documents is practical. It has become. By setting such a bookmark as a noteworthy location in advance, the user can quickly browse a noteworthy location without performing exploratory scrolling.

Here, setting a bookmark interactively while checking a document image is a very time-consuming work, and it is practical to apply it when generating a digitized document from a large number of paper documents. is not. For example, Japanese Patent Laid-Open No. 9-297765 (Patent Document 1) discloses a technique for determining a headline / title from a character size and the number of lines as a method for interactively setting a bookmark. For example, a technique is shown in which a line whose character size is 1.3 times larger than the most frequent character size is used as a headline, and a line whose size is 1.5 times larger is used as a title.
JP-A-9-297765

  However, in Japanese Patent Laid-Open No. 9-297765 (Patent Document 1), it is possible to determine a headline / title, etc. according to the character size, that is, to determine a headline / title, etc. based on whether or not a specific style is used. In the case where there is no specific styling such as character size, the above publication cannot determine headlines, titles, and the like.

  On the other hand, the specific styling such as the character size is not necessarily required for the user, and it may be possible to distinguish them if the same styling is consistently applied to different elements such as headings and titles. It is thought that there are many.

  The present invention has been made to solve the above problems, and provides a document processing apparatus and a document processing method capable of extracting content areas such as headings and titles with high accuracy. Objective.

  A document processing method according to the present invention is a document processing method for generating an electronic document including a document image, the step of extracting at least one content area from the document image, and a document for at least one extracted content area. The method includes obtaining attribute information including position information and character information indicating a position in the image, and generating browsing navigation information for specifying the position of the content area in the document image. The step of generating browsing navigation information corresponds to the step of classifying at least one content area into at least one group based on the character information of the corresponding attribute information, and the content area belonging to each classified group. The method includes a step of calculating a statistic of each classified group based on the position information of the attribute information, and a step of determining each of the groups from a predetermined classification based on the calculation result of the statistic.

Preferably, the predetermined classification corresponds to a title, a heading, a header, a footer, and a text.
Preferably, the character information of the attribute information includes at least one of character size, character color, and character modification.

  Preferably, the document image is divided in units of pages, and the step of calculating the statistic of each classified group includes, for the content area belonging to each group, the position of the content area in the divided page of the document image. The step of calculating an average and a variance value, and the step of calculating the average, minimum, maximum and variance values of the positions defined according to the appearance order of the content area in the entire document image for the content area belonging to each group.

  In particular, the step of generating browsing navigation information further includes the step of calculating the area of the entire document image. The step of determining includes a step of determining a group corresponding to the text based on a ratio of the total area of each group to the area of the entire document image.

  In particular, the step of determining further includes a step of determining a group corresponding to the headline based on a statistic of the position of the group corresponding to the text among the classified groups.

  In particular, the step of determining further includes a step of determining a group corresponding to the title based on the statistics of the positions of the group corresponding to the text and the group corresponding to the heading among the classified groups.

  In particular, the step of generating browsing navigation information further includes a step of calculating an average of the positions of the content areas in the divided pages of the entire document image. The step of determining may include a header or based on a comparison of the live average of the segmented content area of the document image in the classified group and the average of the location of the content area within the segmented page of the entire document image. The method further includes the step of determining a footer equivalent group.

  Preferably, the method further includes a step of generating a document image by reading a document and a step of generating an electronic document by adding browsing navigation information to the document image.

  A document processing apparatus according to the present invention is a document processing apparatus that generates an electronic document including a document image, wherein at least one content area is extracted from the document image, and the position indicating the position of the content area in the document image An acquisition unit that acquires attribute information including information and character information, and an information generation unit that generates browsing navigation information for specifying the position of the content area in the document image are provided. The information generation unit includes classification means for classifying at least one content area into at least one group based on character information of the corresponding attribute information, and the position of the attribute information corresponding to the content area belonging to each classified group Calculation means for calculating a statistic regarding the position of each classified group based on the information, and determination means for determining each of the groups from a predetermined classification based on the calculation result of the statistic.

  According to the document processing method of the present invention, at least one content area is classified into at least one group based on the character information of the corresponding attribute information, and the attribute information corresponding to the content area belonging to each classified group is stored. Based on the position information, a statistic regarding the position of each classified group is calculated. Based on the calculation result of the statistic, each group is determined from a predetermined classification. Accordingly, the content areas can be grouped based on the attribute information, and the content areas such as headings and titles can be extracted with high accuracy in order to determine the group into a predetermined classification based on the statistics regarding the position.

  Embodiments of the present invention will be described in detail with reference to the drawings. Note that the same or corresponding parts in the drawings are denoted by the same reference numerals, and description thereof will not be repeated.

(Embodiment)
(Overall system configuration)
FIG. 1 is a schematic configuration diagram of a system including a document processing apparatus according to an embodiment of the present invention.

  In the present embodiment, an MFP (Multi Function Peripheral) equipped with the document processing apparatus according to the present invention will be typically described. The document processing apparatus according to the present invention can be applied not only to the MFP but also to a copying machine, a facsimile apparatus, a scanner apparatus, and the like.

  Referring to FIG. 1, MFP 1 according to the embodiment of the present invention includes an image reading unit 104 for reading a document 300 and a printing unit 106 for performing a printing process on a paper medium or the like. .

  In particular, MFP 1 according to the embodiment of the present invention obtains a document image by reading document 300 by image reading unit 104 and generates digitized document 400 including the document image. Typically, the electronic document 400 can employ a format such as PDF (Portable Document Format).

  In addition, the MFP 1 extracts the content area included in the document image, acquires attribute information about each content area, and the content in the document image, such as a title / content, for a specific area among the extracted content areas. Browsing navigation information for specifying a headline or the like is generated.

  In this specification, a “content area” is an information resource included in a document, and is a content element (content) in units of lines. In the specification, “browsing navigation information” is information for assisting the user in browsing the document image included in the digitized document, and more specifically, among the content areas included in the document image. This is information for specifying a position where a predetermined object exists. Such browsing navigation information includes “bookmark”, “annotation”, “thread”, “link”, and the like as an example, and includes at least information for specifying the position of the content area in the document image. In the embodiment of the present invention, a configuration using “bookmark” as a representative example of “browsing navigation information” will be described.

  The MFP 1 stores the generated electronic document 400 in its own storage unit (not shown) or transmits it to the personal computers PC1, PC2, PC3 (hereinafter also referred to as “personal computer PC”) via the network. To do.

  As a typical usage pattern, the MFP 1 converts the electronic document 400 from the MFP 1 to the personal computers PC 1 and PC 2 connected to a LAN (Local Area Network) that is a network laid in the same office where the MFP 1 is installed. Is sent directly.

  On the other hand, a server SRV is provided at a connection point between a LAN and a WAN (Wide Area Network). A personal computer PC3 or the like in an office remote from the MFP 1 is connected from the MFP 1 via the server SRV. Thus, the electronic document 400 is transmitted. Here, the server SRV typically includes a mail server, an FTP (File Transfer Protocol) server, a Web server, an SMB server, and the like.

  The image reading unit 104 includes a loading table for setting a document, a document table glass, a conveyance unit that automatically conveys the documents set on the loading table one by one to the document table glass, and a read document. Including a discharge table for discharging. As a result, a plurality of documents can be continuously read and generated as one electronic document 400.

(Schematic configuration of MFP)
FIG. 2 is a block diagram illustrating a schematic configuration in MFP 1 according to the embodiment of the present invention.

  Referring to FIG. 2, MFP 1 according to the embodiment of the present invention includes a control unit 100, a memory unit 102, an image reading unit 104, a printing unit 106, a communication interface unit 108, and a data storage unit 110. Including.

  Control unit 100 is typically composed of an arithmetic device such as a CPU (Central Processing Unit), and implements the document processing method according to the embodiment of the present invention by executing a program.

  The memory unit 102 is typically a volatile storage device such as a DRAM (Dynamic Random Access Memory), and holds a program executed by the control unit 100, data necessary for executing the program, and the like.

  The communication interface unit 108 is typically a part for transmitting and receiving data to and from the personal computer PC via a network (for example, a LAN), and includes, for example, a LAN adapter and driver software for controlling the LAN adapter.

  The print unit 106 is a part for performing print processing, and includes a control device for controlling the operation of each unit in addition to the hardware configuration related to print processing.

  The data storage unit 110 is typically a nonvolatile storage device such as a hard disk device or a flash memory, and stores the digitized document 400 generated by the control unit 100.

  FIG. 3 is a block diagram illustrating a schematic configuration of personal computer PC according to the embodiment of the present invention.

  Referring to FIG. 3, personal computer PC according to the embodiment of the present invention is required for CPU (Central Processing Unit) 201 that executes various programs including an operating system (OS), and for the execution of the programs of CPU 201. A memory unit 213 that temporarily stores data and a hard disk unit (HDD: Hard Disk Drive) 211 that stores a program executed by the CPU 201 in a nonvolatile manner. The hard disk unit 211 stores a browsing application for displaying an electronic document generated by the MFP 1, and such a program is stored in the flexible disk 217 a or the CD-ROM drive 215 by the FDD drive 217 or the CD-ROM drive 215, respectively. It is read from a CD-ROM (Compact Disc-Read Only Memory) 215a or the like.

  The CPU 201 receives an instruction from the user via the input unit 209 including a keyboard and a mouse, and outputs a screen output generated by executing the program to the display unit 205. Further, the CPU 201 acquires a digitized document from the MFP 1 connected to the LAN or WAN via the communication interface unit 207 including a LAN card or the like, or the server device SRV, and stores it in the hard disk unit 211. The above-described units exchange data with each other via the internal bus 203.

  FIG. 4 is a block diagram illustrating a functional configuration in MFP 1 according to the embodiment of the present invention. These functions are mainly realized by the control unit 100 and the memory unit 102 of the MFP 1.

  Referring to FIG. 4, the functional configuration of MFP 1 according to the embodiment of the present invention includes an image reading unit 104, an image preprocessing unit 12, an image buffer unit 13, a compression processing unit 14, and digitized document generation. A unit 15, an image analysis unit 16, a bookmark data generation unit 17, a transmission unit 18, an image processing unit 19, a printing unit 106, and a data storage unit 110 are included.

  The image reading unit 104 reads the document 300 to acquire a document image, and outputs the document image to the image preprocessing unit 12. The image preprocessing unit 12 adjusts the display characteristics of the document image so as to be suitable mainly for display on a personal computer PC or the like. Furthermore, the image preprocessing unit 12 can also remove noise included in the document image. Then, the document image subjected to the image processing by the image preprocessing unit 12 is sent to the image buffer unit 13.

  The image buffer unit 13 is a part for temporarily storing the acquired document image data. The once stored document image is output to the compression processing unit 14, the image analysis unit 16, and the image processing unit 19.

  The compression processing unit 14 compresses the document image output from the image buffer unit 16 and outputs the compressed document image to the digitized document generation unit 15. The degree of compression by this compression process can be changed according to the size of the generated electronic document, the required resolution of the document image, and the compression process is performed by JPEG (Joint Photographic Experts Group). It may be an irreversible conversion. If high resolution is required, the compression process can be omitted.

  The image analysis unit 16 extracts a content area for each page of the document image output from the image buffer unit 13 and acquires attribute information about each extracted content area. Here, the attribute information includes the position in the document image for each content area, the size of the character included in the content area, the color of the character included in the content area, the background color of the content area, and the like. It is. These pieces of attribute information are sent to the bookmark data generation unit 17.

  The bookmark data generation unit 17 generates bookmark data for a specific one of the extracted content regions based on the attribute information of each content region output from the image analysis unit 16. Then, the bookmark data generation unit 17 outputs the generated bookmark data to the digitized document generation unit 15. The digitized document generating unit 15 generates a digitized document by adding bookmark data from the bookmark data generating unit 17 to the document image compressed by the compression processing unit 14.

  Then, the generated electronic document is stored in the data storage unit 110 or output to the transmission unit 18 according to the setting by the user. The transmission unit 18 is realized by the communication interface unit 108, and transmits the digitized document generated by the digitized document generation unit 15 to a personal computer PC or the like via a network such as a LAN.

  On the other hand, the image processing unit 19 converts the document image output from the image buffer unit 13 into an image suitable for the printing operation in the printing unit 106 according to a user operation. Typically, a document image defined by an RGB display system can be converted into image data of a CMY display system suitable for color printing. At this time, color adjustment according to the characteristics of the print unit 106 can also be performed. The printing unit 106 can execute a printing process on a paper medium or the like based on the image data output from the image processing unit 19.

FIG. 5 is an example of a document image according to the embodiment of the present invention.
Referring to FIG. 5, a document image 420 is shown here, and is composed of a text document consisting of three pages.

FIG. 6 is a diagram for explaining the types of content areas in the document image of FIG.
Referring to FIG. 6, here, “header”, “title”, “headline”, “text”, and the like are indicated as the types of content areas.

  Specifically, a “header” is provided at the upper right corner of each page, and is described as “2004/6/8”. Following the “header” on the first page, there is a description of “document sample” which is “title”. This description is underlined with the same font size as that of the “text” and is described in black. In addition, “heading” is described in red with a font size smaller than “text”. The first page is provided with a “headline” described as “full text”, and the second page is provided with a “headline” described as “emperor” and “abandonment of war”. The third page is also provided with a “heading” that describes “National rights and obligations ...” and “National Diet”.

  The body text is assumed to be written in blue with the same font size as the title, underlined.

(Digitized document generation procedure)
FIG. 7 is a flowchart illustrating a specific example of the digitized document generation process according to the embodiment of the present invention.

  The processing shown in the flowchart of FIG. 7 is realized by the control unit 100 reading out and executing a program in the memory unit 102 and controlling each function shown in FIG.

  Referring to FIGS. 4 and 7, first, image reading unit 104 reads document 300 according to a user setting or the like to generate a document image (step S100).

Next, the image preprocessing unit 12 adjusts the generated document image (step S102).
The adjusted document image is stored in the image buffer unit 13.

  Subsequently, the compression processing unit 14 compresses the document image stored in the image buffer unit 13 and outputs it to the digitized document generation unit 15 (step S104).

  On the other hand, the image analysis unit 16 extracts the content area from the document image stored in the image buffer unit 13 in units of lines (step S106).

  Then, the image analysis unit 16 sets a browsing path as a reference for specifying the position of each content area in the document image according to the content area included in the document image of the first page (step S108).

  Further, the image analysis unit 16 acquires a position (“position on the browsing path”) and a position on the page based on the browsing path of each extracted content area (step S110).

  At the same time, the image analysis unit 16 obtains the attribute of each extracted content area (step S112). Specifically, in addition to the “page”, “position on the page”, and “position on the browsing path” of each content area described above, “character size”, “character color (R, G, B)”, “ Get "character modification". These are output to the bookmark data generation unit 17 as attribute information.

  The bookmark data generation unit 17 classifies the entire content area into at least one group using the attribute information of each content area (step S114).

  Thereafter, the bookmark data generation unit 17 calculates the statistics of the content area in the entire document (step S116).

  Next, the bookmark data generation unit 17 calculates the statistics of the content area for each group (step S118).

Next, the type of each group is determined (step S120).
Specifically, the bookmark data generation unit 17 determines the type of each group based on the calculated statistics.

  Then, the bookmark data generation unit 17 generates bookmark data indicating the position in the document image of each content area belonging to the group according to the group determined in this way (step S112).

  Subsequently, the digitized document generation unit 15 generates a digitized document by adding the bookmark data from the bookmark data generation unit 17 to the (compressed) document image from the compression processing unit 14 (step S124). Then, the digitized document generation process ends.

(Content area extraction processing)
FIG. 8 is a diagram for explaining the content region extraction processing in units of rows in step S106 of FIG.

  Referring to FIG. 8, here, the result of executing the content region extraction process in units of lines is shown for the document image 420 described in FIGS. 5 and 6.

  This document image 420 includes page areas 421, 422, and 423, and “title”, “header”, “headline”, and “text” are described in each page area as described above.

  The image analysis unit 16 sequentially extracts a rectangular area including a character string from the document image in units of lines.

  As shown here, a plurality of content areas 430 are extracted in each of the page areas 421, 422, and 423 included in the document image 420. For such extraction processing of the content region 430, a known method disclosed in, for example, Japanese Patent Application Laid-Open No. 09-134406 can be used.

  Here, the type of content area to be extracted can be arbitrarily set in advance. For example, “character string”, “paragraph”, “figure”, “table”, “photograph”, etc. can be extracted. can do. Note that such selection of the type of extraction target can also be realized by using a known technique as described above.

  Here, the configuration for extracting the content area 430 from a horizontally written document will be described. However, the content area 430 can also be extracted from a vertically written document.

  In this case, the image analysis unit 16 extracts each content region 430 by regarding the vertical direction of the paper as the “row direction”. The distinction between “horizontal document” and “vertical document” can be determined based on the density of content elements in the document. Specifically, in general, in “horizontal writing”, content elements are concentrated on the left side of the paper, while the density of content elements on the right side of the paper is low. On the other hand, in the “vertically written document”, the content elements are concentrated on the upper side of the paper, while the density of the content elements on the lower side of the paper is low. Based on such uneven distribution of content elements, “horizontal writing original” and “vertical writing original” can be distinguished.

  The following description will be made by assigning the identification numbers of “row 1” to “row 23” to each content area 430 extracted for convenience, but the image analysis unit 16 does not necessarily have to assign such an identification number. It is also possible to identify (specify) the content area by arranging the content area 430 in the order of appearance (extraction order).

(Browsing path setting process)
FIG. 9 is a diagram for explaining browsing path setting processing in step S108 of FIG.

  Referring to FIG. 9, the image analysis unit 16 sets a browsing path 440 according to the position of the extracted content area 430. More specifically, the image analysis unit 16 sets a viewing path 440 that is on the start point side of the line of the document image and extends in a direction orthogonal to the line. Then, the start point (reference point) of the browsing path 440 is set on the head side of the line arranged in the document image. That is, the browsing path 440 extends in a direction corresponding to the document description order. Typically, in the case of a “horizontal writing document”, as shown in FIG. 9A, a browsing path 440 is set in which the left end of the sheet is directed from above the sheet to below the sheet. The browsing path 440 is set at the same position for each page area included in the document image. The reference point of the browsing path 440 is set on the left side of the page. In the case of “vertical document”, a browsing path is set from the right side to the left side of the paper. In this case, the reference point of the browsing path is set at the upper right of the page. Others are the same as in the case of “horizontal writing original”.

  Here, the browsing path 440 is set to “the starting point side of the line”, and this “starting point side of the line” is determined according to the content area located on the most starting point side in the extracted content area 430. That is, in order to set the viewing path 440, it is necessary to extract the content area 430 included in the document image that is located closest to the starting point. However, when a large number of originals are read by the image reading unit 104 to generate an electronic document, the number of target originals cannot be known in advance. Therefore, if an attempt is made to set the browsing path 440 after extracting the content area 430 included in all pages, the efficiency may decrease. Therefore, the image analysis unit 16 according to the present embodiment sets the browsing path 440 based on the content area 430 included in the page area 421 of the first page of the document image 420. Specifically, the image analysis unit 16 acquires a region 450 surrounding the content region 430 existing in the page region 421, and sets a browsing path 440 based on the region 450.

  In this example, the case where the present invention is applied to a single-column document will be described as an example, but the present invention can also be applied to a multi-column document.

  FIG. 9B is a diagram illustrating a case where a browsing path is set for a two-column original.

  In FIG. 9B, a plurality of content areas 430 are extracted in each of the page areas 424 and 425 included in the document image 420 arranged in two columns.

  Then, areas 453 and 455 surrounding the content area are acquired based on the result of the extraction process of the content area 430 included in the page area 424. The browsing path 440 is set in association with the acquired areas 453 and 455, and the browsing path is set so that the area 453 and the area 455 are connected. For the next page and after, ordering is executed according to the same procedure as the browsing path set for one page.

(Location acquisition processing)
The distance acquisition process based on the browsing path 440 for each content area in step S110 in FIG. 7 will be described with reference to FIG.

  With reference to FIG. 9, in this specification, a line position corresponding to each content area 430 on the browsing path 440 is defined as a “position on the browsing path”. For example, the “position on the browsing path” is a distance indicated by reference numeral 452. As the “position on the browsing path”, an “absolute value” calculated using the starting point of the first page as a reference is used.

  In step S110, for each content area, the position on the page calculated with reference to the starting point of the corresponding page is acquired together with the page number.

  In this manner, the image analysis unit 16 acquires “position on the browsing path” and “position on the page” for each of the extracted content areas 430.

(Other attribute information acquisition processing)
The process of acquiring representative values of “character size”, “character color”, and “character modification” as the attributes of each content area in step S112 in FIG. 8 can be executed using a known character recognition technique or the like. It is. Image analysis unit 16 according to the present embodiment performs character recognition of each content area and acquires “character size”, “character color”, and “character modification”.

  Specifically, a representative value of an attribute value corresponding to a character area included in each content area can be used. The character size can be obtained from the intermediate data of the extracted content area. Alternatively, it can be determined from the height of the extracted content area.

  The character color is a representative value of pixel values other than the background color included in the character region, and the background color is RGB (1, 1, 1) as white, and an average value of pixel values other than the background color is obtained. Is calculated by

  The types of character modification include bold, underline, and enclosure. The type of character modification is identified by using character color density, run length distribution, or the like. Note that these attribute values do not need to be determined as absolute attribute values, and data that can be relatively distinguished can be used.

  Here, when a plurality of types of character sizes, character colors, and the like are included in each content area, it is possible to employ the most frequently used value or an average value for all values.

(Attribute information)
FIG. 10 is a diagram illustrating a specific example of attribute information of each content area acquired from the document image 420 illustrated in FIG.

  Referring to FIG. 10, image analysis unit 16 outputs data as described in data columns 461 to 467 as attribute information for each content area 430 extracted from document image 420.

  Here, the “group ID” stored in the data column 467 in FIG. 10 is not included in the attribute information output by the image analysis unit 16 in order to determine the bookmark data generation unit 17 described later.

(Classification process into groups)
FIG. 11 is a flowchart for explaining the group classification processing in step S114 of FIG. The classification process is executed by the bookmark data generation unit 17.

  Referring to FIG. 11, the group ID table is first initialized (step S11). Then, referring to the position on the browsing path of each content area, the attribute information is acquired in order from the top (step S12).

  Next, referring to each attribute (character size, character color, character modification), it is determined whether there is a table of attributes similar to the group ID table (step S13).

  In step S13, if there is no similar attribute table, a new table is added to the group ID table and each attribute value is stored. And ID of a new table is made into group ID (step S14).

  On the other hand, referring to each attribute in step S13, if there is a table having an attribute similar to the group ID table, the ID of the similar table is set as the group ID (step S15).

  Then, it is determined whether or not a group ID has been assigned to all content areas (step S16). When group IDs are assigned to all content areas, group classification is terminated.

  On the other hand, if group IDs are not assigned to all the content areas in step S16, the process returns to step S12, and the above process is repeated to assign a group ID for the next content area.

  FIG. 12 is an example for explaining a group ID table classified based on the attribute information of the content area in FIG.

  Here, in step S15 of FIG. 11, the group ID table classified in the case where the range in which the character size is ± 1 is similar and the range in which the hue does not change is similar in the character color. It is shown.

  For example, referring to FIG. 11, in step S13, the character size for row 1 is “5”, the character color (R, G, B) is (0, 0, 0) black, and there is no character modification. The group ID number is 1. Similarly, each of the rows 1 to 23 is grouped and given a group ID number.

  In this example, ID numbers 1 to 4 of group IDs are provided, and as an attribute pattern, black and characters with a character size of “5” and a character color (R, G, B) of (0, 0, 0) Group ID number 1 without modification, group ID number 2 when the character size is “10”, the character color (R, G, B) is black (0, 0, 0), and the character modification is underlined The character size is “8”, the character color (R, G, B) is (0.8 to 1, 0, 0) red, the group ID number 3 without character modification, and the character size is “10 to 10”. .5 ", a table of four attribute patterns of group ID number 4 when the character color (R, G, B) is blue (0, 0, 0.5 to 0.8), and there is an underline as character modification It is shown.

  As described above, the bookmark data generation unit 17 classifies each attribute value and determines a group ID by integrating these classification results. In other words, the bookmark data generation unit 17 classifies the content area into at least one group according to processing according to the classification result of each attribute value.

  The grouping process as described above is a process for grouping content areas having a common type (for example, “title”, “headline”, etc.) appearing in the document. In other words, content areas that share the same type are considered to have similar "character size", character color, and character modification, so content areas that have approximate attribute information as described above By grouping each other, it is possible to realize a classification process according to the type in the document.

  Then, as indicated by data 467 in FIG. 10, group IDs are assigned to each of the rows 1 to 23 based on the group ID table.

(Statistics calculation process)
FIG. 13 is a diagram for explaining the distribution of positions on the browsing path for each content area in a document. Here, the horizontal axis indicates the position on the browsing path, and the vertical axis indicates the area. And the case where each content area is plotted according to the group ID number is shown.

  FIG. 14 is a diagram for explaining the distribution of positions on the page for each content region in the document. Here, the horizontal axis indicates the position on the browsing path, and the vertical axis indicates the area. And the case where each content area is plotted according to the group ID number is shown.

  In step S116 of FIG. 7, the bookmark data generation unit 17 calculates the statistics of the entire document based on the attribute information of each content area 430 from the image analysis unit 16. Specifically, the statistic of the position of the content area is calculated based on the data shown in FIGS. Further, statistics such as the number of content areas grouped and the total area are also calculated.

  FIG. 15 is a diagram for explaining the result of calculating the content region statistics for the entire document in step S116.

  Referring to FIG. 15, here, the number of content areas of the entire document, the total area of the number of content areas, the average value and variance value of the position of the content area on the page, and the average value and variance value of the position on the browsing path are as follows. It is shown. As an example, the number of regions is “23”, the total area is “10459”, the average position on the page is “19.7”, the variance is “186.0”, The average value is shown as “69.1” and the variance value is shown as “1638.2”.

  FIG. 16 is a diagram for explaining the result of calculating the statistic of each group ID in step S118.

  Referring to FIG. 16, here, the number of content areas included in each group, the total area of the content areas included in each group, the average value and variance of the positions of the content areas on the page, and on the browsing path An average value and a variance value of the positions are calculated. In the parenthesis for the total area, the ratio to the total area of the entire sentence is shown.

  As an example, the number of areas in the case of group ID number 1 is “3”, the total area is “180”, the average value of the positions on the page is “1.3”, and the variance is “0.2”. The calculated case is shown. Further, the case is shown where the average value of the positions on the browsing path is calculated as “48.8”, the minimum value is “2”, the maximum value is “95”, and the variance value is “1441.6”. Similarly, group ID numbers 2 to 4 are calculated according to the same method.

  Based on these results, the type of each group is determined in step S120 described with reference to FIG.

  Specifically, the determination is made based on the total area, the variance value of the position on the page, and the like, the variance value of the position on the browsing path, and the like.

FIG. 17 is a flowchart for determining the type of each group.
Referring to FIG. 17, the total statistics and the statistics of each group are compared to determine a group corresponding to the text, header, or footer (step S22).

  Specifically, a group corresponding to the text, header, or footer is obtained based on the determination conditions shown in the following table.

  Here, a group having a large value close to the entire value with respect to the total area, the dispersion of the positions on the page, and the dispersion of the positions on the browsing path is determined as the body equivalent group. More specifically, as an example, a group having a value of 50% or more of the total statistic value is set as a text equivalent group. The text equivalent group is considered to have the largest number of areas and the total area in the entire text, considering the general layout. It is thought that there is no bias and the value of dispersion is large. Therefore, a group that satisfies the corresponding condition is determined as a body equivalent group.

  Next, the header of a group whose total area is small compared to the whole value, the variance of the position on the page is extremely smaller than the whole value, and the variance of the position on the browsing path is a large value close to the whole value. Or it determines with a footer equivalent group. The header equivalent group is considered to have a small number of areas and a small total area in consideration of a general layout, and the position on the page may be arranged using a partial area of the upper area for each page. Since there are many, the value of the dispersion | distribution of the position on a page is considered very small. In addition, since the positions on the browsing path are arranged uniformly for each page, it is considered that the value of dispersion becomes large. More specifically, as an example, the total area is 20% or less of the total value, the variance value of the position on the page is 5% or less of the total value, and the position on the browsing path is 50% or more of the total value. Is determined to be a header or footer equivalent group.

  When it is determined as a header or footer equivalent group, whether it is a header equivalent group or a footer equivalent group can be determined by an average value of positions on the page. The header or footer is an additional part in the text. Considering a general layout, the header is located in the uppermost area for each page, and the footer is considered to be located in the lowermost area of the page. Therefore, more specifically, as an example, when the average value of the positions on the page is smaller than the average value of the entire values, it is determined as a header equivalent group. Further, when the average value of the positions on the page is larger than the average value of the whole values, it is determined as a footer equivalent group.

  Referring to FIG. 16, in this example, when the statistic of group ID number 4 is compared with the overall statistic as an example, the total area, the distribution of positions on the page, and the distribution of positions on the browsing path are as follows. Since it has a large value close to the overall value, it can be determined that the group is a group corresponding to the main body.

  Also, in this example, when the statistic of group ID number 1 is compared with the overall statistic as an example, the total area is smaller than the overall value, and the variance of the position on the page is also smaller than the overall value. Since it is extremely small and the dispersion of positions on the browsing path has a large value close to the entire value, it can be determined that the group is equivalent to a header or footer. In this case, since the average value of the positions on the page is smaller than the average value of the whole values, it can be determined as a header equivalent group.

  Next, referring again to FIG. 17, the statistic of the text equivalent group and the statistic of the undetermined group are compared to obtain a headline equivalent group (step S23).

  Specifically, a group corresponding to a headline is obtained based on the determination conditions shown in the following table.

  Here, the total area has a value smaller than the value of the text equivalent group, the average value of the positions on the browsing path has a value close to the value of the text equivalent group, and the variance of the positions on the browsing path is the text. A group having a large value close to the equivalent group value is determined as a headline equivalent group. Considering a general layout, the headline equivalent group is characterized in that the headline has a smaller number of characters and occupies a smaller area than the corresponding part corresponding to the text. Further, since it is arranged near the corresponding part corresponding to the text, the average value of the position on the browsing path is considered to be close to the average value of the position on the browsing path of the text equivalent group. In addition, like the text equivalent group, since they are arranged in a distributed manner, there is no bias in the position on the browsing path, and the value of the variance is considered to be large.

  More specifically, as an example, the total area of undetermined groups is less than the total area of the text equivalent group, and the average value of the positions on the browsing path is close to the value of the text equivalent group. In order to determine whether or not it has, the absolute value of the difference between the average value of the position on the browsing path of the body-equivalent group and the average value of the position on the browsing path of the body-equivalent group is A group that is 30% or less of the average value of the positions on the path, and further that the variance value of the positions on the browsing path to be determined is 70% or more of the variance value of the positions on the browsing path of the body equivalent group Is determined to be a group corresponding to a headline.

  Referring to FIG. 16, in this example, when the statistic of group ID number 3 is compared with the group corresponding to the text as an example, the total area, the average of the positions on the browsing path, and the distribution of the positions on the browsing path are as follows: Since the above conditions are satisfied, it is possible to determine that the group is equivalent to a headline.

  Next, referring again to FIG. 17, the statistics of the headline and body equivalent group are compared with the statistics of the undetermined group, and a group of title candidates is first obtained (step S24).

  Specifically, a group of candidate titles is obtained based on the determination conditions shown in the following table.

  Here, the minimum value among the minimum values of the position on the browsing path of the headline and the body equivalent group is obtained, and an undecided group whose maximum value of the position on the browsing path is larger than that value is determined as the title candidate group. judge. Considering the general layout of the title candidate group, it is considered that the title is often located before the position of the content area of the headline and body equivalent group as the position. It is considered to be smaller than the minimum value of the headline and text equivalent group.

  Referring to FIG. 16, in this example, the statistic of group ID number 2 is found as an example, and when compared with the text equivalent group, the position on the browsing path satisfies the above-described conditions, so that the title candidate group It can be determined that there is.

  Next, referring to FIG. 17 again, next, a group corresponding to a title is determined from among those determined as title candidates. Specifically, first, it is determined whether there are a plurality of title candidate groups (step S25). If there are a plurality of title candidate groups in step S25, the larger total area is determined as a group corresponding to the title (step S27). Considering a general layout, the title is considered to be the largest of the character size and the total area when there are a plurality of title candidate groups.

  On the other hand, if there is only one title candidate group, the title candidate group is determined as a group corresponding to the title (step S26).

  Referring to FIG. 16, in this example, as an example, only group ID number 3 corresponds to a title candidate group, so that it can be determined as a group corresponding to a title.

  Next, referring to FIG. 17 again, further narrowing down is performed using the attribute of each region from each group obtained in the above (step S27). For example, when a predetermined number or more of areas are included in a group corresponding to a headline, some content areas can be deleted from the group. As an example, when a group corresponding to a headline is obtained, if the number of areas is two or more, it is possible to delete other than the one with a large character size. Thus, for example, instead of outputting all the content areas included in the group corresponding to the headline, the number of bookmarks per page can be reduced by selectively outputting a part using the local attributes of the content area. It becomes possible to restrict.

  Based on the information obtained by the above-described processing, the bookmark data generation unit 17 generates the digitized document 400 by adding bookmark data indicating the position of the selected content area in the document image to the document image.

  FIG. 18 is a diagram illustrating an example of a data configuration of the digitized document generated by the digitized document generation unit 15.

  With reference to FIG. 18A, the digitized document includes a header part 402, a main body part 404, a bookmark data part 405, and a footer part 406.

  The header part 402 and the footer part 406 store information about the attributes of the digitized document, such as creation date / time / creator / copyright information.

  The main body 404 stores a document image corresponding to each page. The bookmark data unit 405 stores bookmark data generated by the bookmark data generation unit 17.

  As shown in FIG. 18B, there is a plurality of page data and bookmark data, and a page tree is arranged under a document category via a bookmark hierarchy. Specifically, the order between pages is described in the page tree.

FIG. 19 is a diagram illustrating an example of the data structure of the bookmark data unit 405.
Referring to FIG. 19, bookmark data portion 405 stores position information indicating the position of the content area of the selected group in the document image. In this example, a case will be described in which bookmark data is generated for the title and headline content areas.

  For example, “page 1, (4)”, which is position information stored as “bookmark 1”, has a target content area of the “1” page of the document image and “position on the page” of “4”. It is shown that. The “bookmark 1” corresponds to the title of the group ID number 2.

  In addition, “page 1, (14)”, which is position information stored as “bookmark 2”, is the “1” page of the document image as the target content area, and “position on the page” is “14”. It is shown that. The “bookmark 2” corresponds to the heading of the group ID number 3. It is assumed that bookmark data is similarly generated for other bookmarks.

  Browsing navigation information such as bookmarks can be efficiently generated by grouping and classifying according to each content area included in the document image by the processing as described above.

  In this example, the electronic document 400 has been described with respect to the configuration in which one group is assigned to each of the title, header, footer, heading, and body. However, the same applies to the case where a plurality of groups are assigned. Is possible.

  In the present embodiment, the grouping method using the character size, character color, and character modification type in the region has been described as the content region attribute. However, attribute information for identifying other characters is used. Is also possible.

  For example, it is also possible to use character attribute information such as font identification, indentation position, character alignment such as centering and right alignment, front and rear, left and right margin sizes, area width, height to width ratio, and the like.

  In order to deal with document images having more various styles, it is also possible to identify italics, wide characters, etc. in addition to bold, underline, and enclosure as character modification types and include them in the attribute pattern.

  In the present embodiment, since a digitized document in PDF format in which bookmarks are automatically set at appropriate positions can be obtained by scanning a document, a PDF viewer can be used when browsing the digitized document. The document can be efficiently browsed using the bookmark function.

  In the above-described embodiment, the description has been made on the MFP that generates the PDF-format electronic document with the bookmark set. However, the present invention is not limited to the MFP, and the bookmark / title area extracted from the document image is used. The same applies to other devices that use the function.

  In addition, a method for selectively performing character recognition processing on a title / heading area and generating a character code as a search key for a document image, or combining an image corresponding to the title / heading area, a table of contents or a summary It is also applicable to the method of generating With this method, it is possible to extract appropriate areas such as titles and headings for document images of various styles, and it is possible to generate useful search keys, a table of contents, a summary, and the like.

  In the above-described embodiment, the case where the process according to the present invention is executed by the MFP 1 has been described. However, the above process may be executed by a computer having an image reading function for reading the document 300. In this case, a program for executing the processing function shown in FIG. 4 for causing the computer to function as a document processing apparatus can be provided. Such a program is stored in a computer-readable storage medium such as a flexible disk attached to the computer, a CD-ROM (Compact Disk-Read Only Memory), a ROM (Read Only Memory), a RAM (Random Access Memory), and a memory card. And stored as a program product. Alternatively, the program can be provided by being stored in a storage medium such as a hard disk built in the computer. A program can also be provided by downloading via a network.

  Alternatively, the image reading function may be realized by another device or a computer, and the generated document image may be received and only the browsing navigation information may be generated according to the above processing. In addition, the configuration in which the document image and the browsing navigation information are included in the same digitized document has been illustrated, but the browsing navigation information may not necessarily be added to the same digitized document, and is output as a separate file. May be.

  The program according to the present invention is a program module that is provided as a part of a computer operating system (OS) and calls necessary modules in a predetermined arrangement at a predetermined timing to execute processing. Also good. In that case, the program itself does not include the module, and the process is executed in cooperation with the OS. A program that does not include such a module can also be included in the program according to the present invention.

  The program according to the present invention may be provided by being incorporated in a part of another program. Even in this case, the program itself does not include the module included in the other program, and the process is executed in cooperation with the other program. Such a program incorporated in another program can also be included in the program according to the present invention.

  The provided program product is installed in a program storage unit such as a hard disk and executed. Note that the program product includes the program itself and a storage medium in which the program is stored.

  The embodiment disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

1 is a schematic configuration diagram of a system including a document processing device according to an embodiment of the present invention. FIG. 2 is a block diagram illustrating a schematic configuration in MFP 1 according to the embodiment of the present invention. It is a block diagram illustrating a schematic configuration of a personal computer PC according to an embodiment of the present invention. It is a block diagram illustrating a functional configuration in MFP 1 according to the embodiment of the present invention. It is an example figure of the document image according to embodiment of this invention. It is a figure explaining the kind of content area in the document image of FIG. It is a flowchart explaining the specific example of the production | generation process of the digitized document according to embodiment of this invention. It is a figure for demonstrating the extraction process of the content area in the line unit in step S106 of FIG. It is a figure for demonstrating the setting process of the browsing path in step S108 of FIG. It is a figure explaining the specific example of the attribute information of each content area acquired from the document image 420 shown in FIG. It is a flowchart explaining the classification process to the group in step S114 of FIG. It is an example figure explaining the group ID table classified based on the attribute information of the contents area of FIG. It is a figure explaining the distribution of the position on the browsing path about each content area | region in a document. It is a figure explaining distribution of the position on a page about each contents field in a document. It is a figure explaining the result of having calculated the statistics of the content area about the whole document in Step S116. It is a figure explaining the result of having calculated the statistics of each group ID in Step S118. It is a flowchart which determines the kind of each group. It is a figure which shows an example of the data structure of the digitized document which the digitized document production | generation part 15 produces | generates. It is a figure which shows an example of the data structure of the bookmark data part.

Explanation of symbols

  1 MFP, 12 image preprocessing unit, 13 image buffer unit, 14 compression processing unit, 15 digitized document generation unit, 16 image analysis unit, 17 bookmark data generation unit, 18 transmission unit, 19 image processing unit, 100 control unit, 102 memory unit, 104 image reading unit, 106 print unit, 108, 207 communication interface unit, 110 data storage unit, 201 CPU, 203 internal bus, 205 display unit, 209 input unit, 211 HDD, 213 memory unit, 215 CD- ROM drive, 217 FDD drive.

Claims (10)

A document processing method for generating an electronic document including a document image,
Extracting at least one content region from the document image;
Obtaining attribute information including position information and character information indicating a position in the document image for the extracted at least one content area;
Generating browsing navigation information for specifying the position of the content area in the document image,
The step of generating the browsing navigation information includes:
Classifying the at least one content area into at least one group based on character information of corresponding attribute information;
Calculating a statistic of each classified group based on position information of the attribute information corresponding to the content area belonging to each classified group;
And a step of determining each group from a predetermined classification based on the calculation result of the statistic.   The document processing method according to claim 1, wherein the predetermined classification corresponds to a title, a heading, a header, a footer, and a body.   The document processing method according to claim 1, wherein the character information of the attribute information includes at least one of a character size, a character color, and character modification. The document image is divided into page units,
The step of calculating the statistic of each classified group includes:
For the content areas belonging to each group, calculating an average and a variance value of the positions of the content areas in the divided pages of the document image;
The document processing method according to claim 1, further comprising: calculating an average, minimum, maximum, and variance value of positions defined according to the appearance order of the content areas in the entire document image for the content areas belonging to each group. The step of generating the browsing navigation information further includes a step of calculating an area of the entire document image,
5. The document processing method according to claim 4, wherein the determining step includes a step of determining a group corresponding to the text based on a ratio of a total area of each group to an area of the entire document image.   The document processing method according to claim 5, wherein the determining step further includes a step of determining a group corresponding to a headline based on a statistic of a position of a group corresponding to the text among the classified groups.   The document processing method according to claim 6, wherein the determining step further includes a step of determining a group corresponding to a title based on a statistic of a position of a group corresponding to a text and a group corresponding to a heading among classified groups. The step of generating the browsing navigation information further includes calculating an average of the positions of the content areas in the divided pages of the entire document image,
The step of determining is based on a comparison of a living average of the segmented content region of the document image in the classified group with an average of the location of the content region within the segmented page of the entire document image. 5. The document processing method according to claim 4, further comprising a step of determining a group corresponding to a header or footer. Generating the document image by reading a document;
The document processing method according to claim 1, further comprising: generating the digitized document by adding the browsing navigation information to the document image. A document processing apparatus for generating an electronic document including a document image,
An acquisition unit that extracts at least one content area from the document image and acquires attribute information including position information and character information indicating a position of the content area in the document image;
An information generation unit that generates browsing navigation information for specifying the position of the content area in the document image;
The information generator is
Classification means for classifying the at least one content area into at least one group based on character information of corresponding attribute information;
Calculation means for calculating a statistic of each classified group based on the position information of the attribute information corresponding to the content area belonging to each classified group;
A document processing apparatus including determination means for determining each of the groups from a predetermined classification based on the calculation result of the statistics.

Images