JP2008040753A - Image processor and method, program and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve the visibility of a document image on the list display of retrieved results, to make it possible to search a retrieval object image on the list display and to improve operability of keyword retrieval in a document image DB. <P>SOLUTION: When a server unit 110 receives a retrieval key, a keyword retrieval processing part 119 performs keyword retrieval processing to text information inside an image information DB 117. A display screen control processing part 121 displays the thumbnail, text information and representative partial image of a hit page. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a search result display technique in text search, and relates to an image processing apparatus, method, and program having a search function for performing a text search according to a search keyword input by a user, and a display screen creation function for displaying the search result. The present invention relates to a technique suitable for, for example, a multifunction peripheral of a copying machine, a file server, and an image processing program.

  There are devices such as electronic filing that digitize paper documents using an input device such as a scanner, but they have been used exclusively for business purposes dealing with a large amount of paper documents. In recent years, MFPs (Multi Function Printers) equipped with lower scanner prices and scanning functions have become popular, and the legislation such as the e-document method has been recognized for its good handling and convenience in general offices. Opportunities to scan and digitize documents are increasing. In addition, the use of an image DB for centralizing and managing digitized document image data, photographic image data, document data created by an application such as a PC as a database (hereinafter referred to as DB) is also increasing. For example, even if an original paper document is stored, an image DB may be constructed for ease of management and search.

  Such image DBs vary from large-scale ones where a server device is installed and accessed by a large number of people to personal uses for constructing a DB in a personal PC. Recent MFPs have a function of storing documents in a built-in HDD, and there is an example in which an image DB is constructed based on the MFP. Some image DBs have a search function for searching a desired image from a large number of images. Currently, the mainstream search function generally performs full-text search or synonym or concept search using a keyword recognition result obtained from OCR (Optical Character Reader) processing, text information extracted from document data, or the like as a keyword.

The search result of the document image searched in this way is usually presented to the user via some display means. There are the following three methods for such a display method (in this case, not a hardware meaning but a software meaning for creating a display screen).
Method 1: A method of displaying document information such as document file names and creation dates in a text format. Method 2: A method of displaying a thumbnail list using reduced document image data hitting a keyword. Method 3: A keyword. Method of displaying hit surrounding text as text information Method 1 is a method used in a device having a small display screen size and low processing capability, and method 2 is used in a current document image DB or the like. Method 3 is a method used for Internet search or the like.

  In the conventional display method described above, it is difficult for the user to find a desired document image only from the search result screen. As a result, each document image data hit in the search is displayed using a screen such as a viewer, and the search is performed. There was a problem that the target image had to be searched and the operability during the search was poor.

  For example, in the method 1, since the document cannot be searched unless the document information such as the file name and the creation date is known in advance, if the user does not know the information, it is necessary to check with the viewer. In the method 2, search is possible if the document image data has a characteristic overview. However, when there are many hit document images having a similar overview, for example, a fixed form such as a paper, a form, a patent, or the like. When a large number of document images close to a fixed form are hit, it is necessary to confirm with a viewer after all. Furthermore, in method 3, since there is no information for recognizing the appearance of images such as thumbnails, it is difficult to make an intuitive judgment, and it is necessary to read each piece of text information, and the operation is troublesome. As a result, it is impossible to determine whether the image is an image. Eventually, there are many cases where the image is checked by a viewer (the list becomes worse if more text information is displayed). The above problem becomes more prominent as the scale of the image DB increases and the number of hits by keyword search increases.

  Therefore, there is a search system disclosed in Patent Document 1 for such a problem. In Patent Document 1, by generating and displaying a reduced simplified image characterized so that a text object that matches a search key can be identified from other parts, the appearance frequency of the text object can be visually recognized at a glance, Further, it is possible to infer how the object is used in the document from the appearance position of the object in the paragraph, and the document content can be determined.

JP 2004-157668 A

  However, in the above method, the outline of the document image and the position of the keyword can be understood, but it is difficult to grasp the contents of the document image. Generally, a document image search is roughly divided into a case where a user wants to search for a document and searching for the known document and a case of searching for an unknown document in order to obtain information. Cannot be used to retrieve unknown documents. Also, for example, even if a page of a document image composed mostly of characters is displayed by the method described above, it is difficult to determine whether or not it is a search target image.

The present invention has been made in view of the above problems,
An object of the present invention is to provide a document image on a list display of a search result in an image processing apparatus having a function of performing a keyword search of a document image DB based on a keyword input by a user and generating a display screen of the search result. To provide an image processing apparatus, method, program, and recording medium that can improve the visibility of images, search for a search target image only on a list display, and improve the operability of keyword search in the document image DB. is there.

  The present invention is an image processing apparatus having a function of searching document image data stored in an image database, receiving an image dividing means for dividing a document image into partial images, a search key based on a keyword, and the document image A keyword search means for searching for each page, a display screen control means for generating a display screen for displaying one or more document images hit by the keyword search, and a partial image selection means for selecting the divided partial images. The display screen includes a thumbnail showing the overview of the page, text information including the hit keyword, and a partial image selected by the partial image selection means for each page. And

  According to the present invention, by simultaneously displaying the thumbnail of the document image on the search result list display, the text information that hits the search key, and the representative partial image, an overview of the image can be easily grasped by the thumbnail, and the text information The document contents before and after the search key can be easily grasped, and the visibility of details of representative partial images that cannot be confirmed with thumbnails by the representative partial images has been improved, so a known document image can be searched by relying on ambiguous memory. Even when searching for a document image to obtain unknown information, it is possible to help understand the content of the document image that has been hit by the search, and to search for the search target image only on the list display. Improved search operability.

  According to the present invention, when the display partial image is selected, the partial image indicating the title portion that is highly likely to express the sentence content is selected, so that the document content can be easily understood.

  According to the present invention, when selecting a display partial image, a partial image such as a photograph, a figure, or a table other than the body area is selected, so that the contents of the document can be grasped intuitively.

  According to the present invention, since the partial image hit with the search key is selected as the representative partial image, it is easy to grasp which portion in the document image the hit portion of the search key corresponds to by comparing with the thumbnail. it can. In addition, since the outline of the characters (font style and font size) can be understood on the actual image, the document contents can be easily grasped.

  According to the present invention, when selecting a display partial image, since a partial image in the vicinity of the partial image hit in the keyword search is selected, a partial image related to the search keyword can be selected, and a document image based on the search keyword can be selected. Understanding of the contents becomes easier.

  According to the present invention, when selecting a representative partial image, since a partial image of a large size is selected, it is possible to select a partial image that directly represents the contents of the document image and to facilitate understanding of the contents of the document image. .

  According to the present invention, a keyword indicating an object such as a figure, a photograph, or a table included in a partial image whose search key is hit is searched, and a representative partial image is selected based on the keyword. A partial image can be selected, and the contents of the document image can be easily understood by the search keyword.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Example 1:
FIG. 1 shows a system configuration of Embodiment 1 of the present invention. In FIG. 1, reference numeral 100 denotes a client device such as a personal computer (hereinafter referred to as a PC), a mobile terminal such as a PDA or a mobile phone. 101 is a display device such as a monitor, 102 is an application program that interprets user instructions, communicates with the server 110, and controls the display device 101, 103 is an input device such as a keyboard and mouse that is an instruction input means from the user, 104 Is an external communication path such as a LAN or the Internet.

  110 is a server device that performs image classification according to a command from the client and outputs the classification result to the client 100, 111 is an interface with the external communication path 104 (hereinafter referred to as I / F), and 112 is a registered image to be registered in the image DB 113. Data, 113 is a thumbnail generation processing unit that generates a thumbnail image by scaling the registered image 112 to a predetermined size or less, 114 performs layout analysis processing of the registered image 112, and divides the image into partial images (hereinafter referred to as objects) 115 is an attribute identification processing unit that identifies the attribute of the object generated by the image division processing unit 114, 116 is a character code extraction processing unit that extracts a character code from the object image, and 117 is Image information for storing information for each document image data registered in the image DB 120 It is a B. The information includes, for example, the file name, creation date, image data, thumbnail image data, and object image data associated with the registered image data (the association information is, for example, specific to each data when stored in the image DB 120) ID, file name, etc.), attributes for each object, position information of the object in the page, text information (character code) included in the object, and the like.

  118 is a search key composed of text information designated and input by the client 100, 119 is a keyword search for searching for an object or document image corresponding to the search key 118 from information stored in the object information DB, and outputting the search result. The processing unit 120 is the image data of the registered image 112, the thumbnail image data of the registered image 112, the image DB that stores the object image of the registered image 112 generated by the image division processing unit 114, and 121 is the search of the keyword search processing unit 119. A display screen control processing unit that generates a display screen to be displayed on the client 100 according to the result and a selection result of the object selection processing unit, 122 is display screen data displayed on the display device 101 of the client 100, and 123 is a keyword Search by search processing unit 119 An object selection processing unit for selecting the object for display to the client 100 in accordance with the result. The dotted line in the figure represents the data flow at the time of image registration, and the solid line represents the data flow when generating a display screen by performing a keyword search process.

  FIG. 2 shows the configuration of the server device 110. In FIG. 2, 201 is a CPU that performs calculations and processing according to a program, 202 is a volatile memory used as a work area for temporarily storing and holding data such as program code and image code data, 203 Is a hard disk (hereinafter referred to as HDD) for storing and accumulating image data, programs, etc., and holds an image DB 120 and an image information DB 117. Reference numeral 204 denotes a video memory serving as a data buffer for display on the monitor 205. The image data written in the video memory 204 is periodically displayed on the monitor 205. Reference numeral 206 denotes an input device such as a mouse or a keyboard, 207 denotes an external I / F that transmits and receives data via the external communication path 104 such as the Internet or a LAN, and 208 denotes a bus that connects each component.

  In the present embodiment, an example in which the server device 110 is configured by a server computer and processing such as image search is realized by software. That is, the processing in the server is realized by an application program (not shown). The embodiment of the present invention is not limited to this, and may be configured such that processing is performed by hardware in an apparatus such as an MFP. Also, for example, a single PC or MFP can be used without adopting a server or client configuration. 1 may be configured in such a device.

  Hereinafter, an operation at the time of image registration and display of a search result at the time of an image search operation in the present embodiment will be described.

  FIG. 3 is a flowchart of the image registration operation according to the first embodiment. The image registration operation will be described with reference to FIG. 1 (broken lines indicate the operation during registration) and FIG.

  In step S001, the user instructs the server apparatus 110 to register image data and the registered image data 112 to be registered from the client apparatus 100 via the application program 102.

  In step S002, the registered image data 112 is input to the server apparatus 110 along with file information such as a file name and a creation date via the external communication path 104, and an ID number is assigned to the image DB 120 via the external I / F 111. Registered. At the same time, the thumbnail generation processing unit 113 performs scaling processing on the registered image 112 to generate a thumbnail image having a predetermined size or less, and registers the image DB 120 with an ID number. When the registered image data 112 is image data having a plurality of pages, thumbnails are generated in units of pages.

  In step S003, the registered image data 112 is input to the image division processing unit 114, and is divided into regions for the same attribute by a known layout analysis process to generate object image data. At this time, the coordinate data of the object of the image data is also extracted. The generated object image data is registered in the image DB 120 with an ID number assigned to each object image data, and is output to the attribute identification processing unit 115.

  In step S004, the generated object image data is input to the attribute identification processing unit 115 for each object image, and the attribute is identified for each object by a known layout analysis process.

  Note that the layout analysis processing in steps S003 and S004 described above is used in the pre-processing of the OCR processing, and there are various methods, and these known methods may be used. For example, the background color of the document image is specified, pixels other than the background region are extracted from the document image using the background color, the connected components are generated by integrating the pixels, and at least the shape feature is used for the connected components. There is a technique for classifying a character area and a photographic area by classifying them into predetermined areas (see Japanese Patent Laid-Open No. 2001-297303). As another example of character region identification, there is a technique for identifying a character region using a circumscribed rectangle after performing an adaptive binarization process (see JP-A-7-73271). ). Furthermore, the adjacent relationship of the black area of the input image is analyzed and divided into rectangles (image division processing), and the character, photo, figure, and table areas of the input image are based on the size of the rectangle and the distribution density of the black area. There is also a technique (attribute identification process) for identifying (see Japanese Patent Application Laid-Open No. 7-221968).

  By using (or combining) the above-described conventional techniques, it is possible to divide an image area (objectification) for each attribute such as a character area, a photographic area, a graphic area, and a table area, and determine the attribute. At this time, the title area is identified based on the position and size of the character area, the size of the character, etc., and is determined separately from the character area.

  In addition, as another example of attribute identification of the divided object, for example, a neural network in which a histogram of a divided area, a frequency feature amount, etc. are acquired and a relationship between the feature amount and the attribute is learned in advance is used. A pattern recognition method such as a support vector machine may be used. Further, it is more preferable to perform pre-processing such as skew correction and back-off removal on the input image for the purpose of improving the accuracy before layout analysis processing.

  FIG. 4B shows an example of the result of layout analysis performed on the document shown in FIG.

  In step S005, the object image data is input to the character code extraction processing unit 116, and if a character is included in the object, the character code is extracted by performing character recognition by performing OCR processing or the like. Note that, when the input registered image data is bitmap data such as a scanned image, OCR processing is required. However, in the case of document data including a character code, the character code from the data may be extracted as it is.

In step S006, the following image information data generated and extracted in steps S002 to S005 is registered in the image information DB 117.
・ File name, creation date ・ Image data ID
・ Thumbnail image data ID
-Object image data ID
-Object coordinates-Object attributes-Text information extracted from the object (by character code) The image information DB 117 uses the general RDB (relational database) to register, manage, search, etc. the above information Can be realized easily. Further, if the image DB 120 and the image information DB 117 satisfy the above functions, a hierarchical data structure or the like may be constructed and stored in the same DB using a language such as XML (extensible Markup Language), for example. These may be stored as different DBs on different servers. As for image registration, image data may be registered in the server apparatus 110 directly from an image input device such as a scanner or a digital camera.

  FIG. 5 is a flowchart of an image search operation in the first embodiment. The image search operation will be described with reference to FIGS.

  In step S <b> 101, the user uses the application program 102 in the client device 100 to specify a search key and instruct the server device 110 to perform an image search. The search key here is assumed to be a keyword such as a word. The instruction means at this time displays a search key designation screen as shown in FIG. 6 on the display device 101 of the client apparatus 100, for example. In FIG. 6, 301 is a keyword input window, 302 is a search button for instructing search, and 303 is a cancel button for canceling search.

  The user inputs a search keyword into the keyword input window 301 using the keyboard of the input device 103 or the like, uses a pointing device such as a mouse of the input device 103, and clicks the search button 302, thereby causing the server 110 to enter. An image classification instruction is transferred via the external communication path 104.

  In step S102, when the server apparatus 110 receives the search key together with the search instruction, the keyword search processing unit 119 performs keyword search processing on the text information of the object stored in the image information DB 117. The keyword search process consists of a method that searches only for a keyword that exactly matches the keyword specified by the search key, a method that searches for synonyms of keywords, associated words, etc., and a natural sentence as a search key. There is a method of searching using the extracted word as a key, but any method may be used in the present invention, and the client device 100 may be configured to designate them.

The keyword search processing unit 119 outputs the following information regarding the object hit in the search from the image information DB 117 to the display screen control processing unit 121.
・ File name, creation date ・ Image data ID
・ Thumbnail image data ID
-Object image data ID
-Object coordinates-Object attribute-Text information extracted from object (by character code) In step S103, the object selection processing unit 123 makes a display target page (keyword search hit) according to the search result of the keyword search processing unit 119. Select the object to display from the objects included in the page containing the object. Details of the selection method will be described later.

  In step S104, the display screen control processing unit 121 determines the layout of the search result display screen 122 according to the search result of the keyword search processing unit 119 and the selection result of the object selection processing unit 123. Next, the image data to be displayed, the thumbnail of the image data, and the object image data are input from the image DB 120, and the text information hit by the keyword search is input from the image information DB 117, and the search result display screen 122 is generated. / F111 transmits to the client 100 via the external communication path 104.

  There are various methods for creating the display screen and for communication between server clients as described above. As a commonly used method, the World Wide Web-based technology is used with the server device 110 as a Web server. This can be realized. The display screen 122 is described in HTML (Hyper Text Markup Language), and the application 102 may use a general Web browser.

  FIG. 7 shows an example of a search result display screen. The display screen in FIG. 7 is referred to as a page detail display screen. In FIG. 7, 311 is a radio button for instructing switching of the display screen to the page detail display screen, 312 is a radio button for instructing general thumbnail display, 313 is a frame for displaying search results, 314 is a page (unit ) Display area 315 is a slider for scrolling the frame.

  As described above, the detailed display screen of the first embodiment is an example in which the display area 314 is provided for each page hit in the search, and the search target image is searched from the result display screen of the keyword search while scrolling the display area 314 with the slider 315. Show.

  FIG. 8 shows an example in which the page search area 314 displays the search result using the “attribute” of the image shown in FIG. 4 as a keyword. In FIG. 8, 321 is the file name, 322 is the thumbnail image of the corresponding page of the image data hit in the search, 323 is the title extracted from the image data, 324 is the selected representative object, 325 is the text data hit in the keyword search It is.

  This example shows an example in which the text data 325 is character code data (hereinafter abbreviated as text data. Bitmap character images are referred to as character images). As a result, the character code is interpreted by the application 102 of the client 100, and the user can read the character code on the display screen. This part may be displayed by cutting out the character image of the necessary part of the image data, but the size is not uniform, and since it is the image data, the operation of making the appearance easy for the user to read is complicated. Text data is better because of its disadvantages. In text data, character decoration operations such as emphasizing or coloring a keyword portion are also easy. Further, a link of actual object image data may be attached to the text portion, and the object image data may be displayed when clicked with a pointing device such as a mouse.

  The first embodiment shows a display form in units of pages. However, in a search using a keyword, there are many cases where a page unit is a search target. For example, when a user finally needs a document consisting of a plurality of pages including a search target page, it is only necessary to first search for a search target image in units of pages and reach a document from the page. There is no problem with the method described above. If the user needs a document consisting of a plurality of pages directly from a keyword search, the thumbnail displayed on the thumbnail image 322 may be used as the top page thumbnail, or the top page thumbnails may be displayed side by side.

  In step S <b> 105, the client apparatus 100 displays the display screen 122 on the display device 101.

Next, the operation of the object selection processing unit will be described. In the first embodiment, an object is selected under the following conditions. FIG. 9 shows a flowchart of the object selection process.
If a title exists, a title object is selected (step S201).
One object closest to the object hit by the keyword is selected from objects other than the body text and title object, such as graphics, photos, and tables (steps S202 and S203).

  For example, when “claim” is a keyword, in FIG. 4B, the object of body 1 is the hit object, and the graphic object of the nearest object is selected.

  In the first embodiment, an example in which there is one selection object other than the title is shown, but a plurality of selection objects may be selected. In addition, when selecting nearby objects, text information obtained from nearby text-based objects is associated with graphics, photos, and table objects in advance, and these objects are automatically selected by keyword search. Also good. In the first embodiment, an image division method for dividing an image area into attribute units is shown. For example, as shown by a broken line in FIG. 10, an image is equally divided and the divided area is displayed as a representative object. Also good. In this case, since the display screen as shown in FIG. 11 is obtained, the accuracy of accurately presenting the contents of the document image is reduced, but the processing of the server device is reduced because the attribute need not be determined.

  As described above, according to the present embodiment, in the image processing apparatus having the function of performing a keyword search of the document image DB based on the keyword input by the user and generating a display screen of the search result, By displaying thumbnails of document images on the list display, text information that hits the search key, and representative objects at the same time, the overview of the image can be easily grasped by thumbnails, and the document contents before and after the search key can be easily determined by text information. In addition, the visibility of the details of representative objects that cannot be confirmed with thumbnails is improved by the representative object, so when searching for a known document image relying on ambiguous memory, or to obtain unknown information Even when searching for document images, it helps to understand the contents of document images that have been found in the search, and Search target image it is possible to find out in the body, thereby improving the operability of the keyword search in the document image DB.

  In addition, when selecting a representative object, a title object that is highly likely to express the content of the sentence is selected, so that it is easy to understand the document content. In addition, when selecting a representative object, a region such as a photo, figure, or table other than the body region is selected, so that the contents of the document can be grasped intuitively. In addition, when selecting a representative object, an object in the vicinity of the object hit in the keyword search is selected, so an object related to the search keyword can be selected, and the contents of the document image can be easily understood by the search keyword.

Example 2:
The second embodiment is different from the first embodiment in object selection means. Other components are the same as those in the first embodiment.

In the second embodiment, an object is selected under the following conditions. FIG. 13 is a flowchart of object selection processing according to the second embodiment.
If a title exists, a title object is selected (step S301).
Select an object whose search key is hit (step S302).
One object of the maximum size is selected from objects other than the body text and title object, such as graphics, photos, and tables (step S303).

  An example of the processing result of the second embodiment for the image of FIG. 4 is shown in FIG. For example, when “claim” is a keyword, in FIG. 4B, the object of body 1 is the hit object, and as shown in FIG. 12, the hit object 326 is selected as the image data. Also, the maximum size photo object 324 is selected.

  As described above, according to the present embodiment, since the object whose search key is hit is selected as the representative display object, the portion where the search key is hit by comparing with the thumbnail is the part in the document image. You can easily grasp whether it corresponds. In addition, an overview of characters on the actual image (font style and font size) can be easily understood. Further, since the representative object is selected according to the size of the object, an object that directly represents the contents of the document image can be selected, and the contents of the document image can be easily understood.

Example 3:
The third embodiment is different from the first embodiment in object selection means. Other components are the same as those in the first embodiment.

In the third embodiment, an object is selected under the following conditions. FIG. 14 is a flowchart illustrating object selection processing according to the third embodiment.
If a title exists, a title object is selected (step S401).
A keyword indicating an object such as a figure, a photograph, a table or the like such as “table”, “FIG”, “figure”, “photograph”, or the like is searched from the text information of the object hit by the search key (step S403). (Yes in S404), keyword search is performed with a keyword hit for the caption of an object such as a graphic, photo, or table (in this example, it is included in the graphic, photo, or table object), and the object Is selected (step S407). The example of the processing result is the same as that in the first embodiment (provided that the “text” keyword is present in the text 3).

  As described above, according to the present embodiment, a keyword indicating an object such as a figure, a photograph, or a table included in an object whose search key is hit is searched, and a representative object is selected based on the keyword. The object related to the search keyword can be selected, and the contents of the document image can be easily understood by the search keyword.

1 shows a system configuration of an embodiment of the present invention. The structure of a server apparatus is shown. 2 is a flowchart illustrating an image registration operation according to the first exemplary embodiment. An example of image segmentation and attribute identification results is shown. 2 is a flowchart of an image search operation according to the first embodiment. An example of a search key designation screen is shown. An example of a thumbnail list display screen is shown. The example of the page display area detail screen of Example 1 is shown. 3 is a flowchart of object selection processing according to the first embodiment. Another method of image segmentation is shown. The example of the display screen by the division | segmentation method of FIG. 10 is shown. The example of the page display area detail screen of Example 2 is shown. 10 is a flowchart of object selection processing according to the second embodiment. 10 is a flowchart of object selection processing according to the third embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Client apparatus 101 Display device 102 Application program 103 Input device 104 External communication path 110 Server apparatus 111 Interface 112 Registered image data 113 Thumbnail generation process part 114 Image division process part 115 Attribute identification process part 116 Character code extraction process part 117 Image information DB
118 Search Key 119 Keyword Search Processing Unit 120 Image DB
121 Display Screen Control Processing Unit 122 Display Screen Data 123 Object Selection Processing Unit

Claims (18)

  An image processing apparatus having a function of searching document image data stored in an image database, receiving an image dividing unit for dividing a document image into partial images, and a search key based on a keyword, A keyword search means for searching; a display screen control means for generating a display screen for displaying one or more document images hit by the keyword search; and a partial image selection means for selecting the partial image obtained by dividing the image. An image processing apparatus characterized in that a screen arranges a thumbnail indicating an overview of a page, text information including a hit keyword, and a partial image selected by the partial image selection unit for each page.   The image processing apparatus according to claim 1, further comprising an attribute identification unit that identifies an attribute of the partial image, wherein the partial image selection unit selects a partial image according to the identification result of the attribute identification unit.   The image processing apparatus according to claim 2, wherein the attribute selected by the partial image selection unit is a page title.   3. The image processing apparatus according to claim 2, wherein the attribute selected by the partial image selection means is one or more of a photo area, a graphic area, and a table area.   The image processing apparatus according to claim 2, wherein the attribute selected by the partial image selection unit is an area including a hit keyword.   The image processing apparatus according to claim 1, wherein the partial image selection unit selects a partial image according to a distance from the partial image including the hit keyword.   The image processing apparatus according to claim 1, wherein the partial image selection unit selects a partial image according to a size of the partial image.   The image processing apparatus according to claim 1, wherein the partial image selection unit selects a partial image according to character information included in the partial image including the hit keyword.   An image processing method having a function of searching document image data stored in an image database, an image dividing step for dividing a document image into partial images, and a search key based on a keyword are received, and the document image is page by page A keyword search step for searching; a display screen control step for generating a display screen for displaying one or more document images hit in the keyword search; and a partial image selection step for selecting the partial image obtained by dividing the image. An image processing method characterized in that the screen arranges a thumbnail showing an overview of a page, text information including a hit keyword, and a partial image selected in the partial image selection step for each page.   The image processing method according to claim 9, further comprising an attribute identification step that identifies an attribute of the partial image, wherein the partial image selection step selects a partial image according to an identification result of the attribute identification step.   The image processing method according to claim 10, wherein the attribute selected in the partial image selection step is a page title.   The image processing method according to claim 10, wherein the attribute selected in the partial image selection step is one or more of a photograph area, a graphic area, and a table area.   The image processing method according to claim 10, wherein the attribute selected in the partial image selection step is an area including a hit keyword.   The image processing method according to claim 9, wherein the partial image selection step selects a partial image according to a distance from the partial image including the hit keyword.   The image processing method according to claim 9, wherein the partial image selection step selects a partial image according to a size of the partial image.   The image processing method according to claim 9, wherein the partial image selection step selects a partial image according to character information included in the partial image including the hit keyword.   A program for causing a computer to implement the image processing method according to any one of claims 9 to 16.   The computer-readable recording medium which recorded the program for making a computer implement | achieve the image processing method of any one of Claims 9 thru | or 16.

Images