JP2004265433A - Construction method of image in image storage device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To easily search for target data to construct them as an image by using a plurality of pieces of data having different data types as search keys. <P>SOLUTION: This image storage device is provided with a part data storage part 1 which stores a plurality of objects as part data capable of raster-drawing on an image, and a relation storage part 3 storing relation information which is constructed of a plurality of objects and image processing operation carried out for drawing a plurality of objects to raster images, for each image. In this image construction method in the image storage device, the part data necessary for the construction of an image are searched from the part data storage part 1 according to desired image relation information stored in the relation storage part 3, and the image is constructed by performing image processing operation, which is defined by the relation information on the searched part data. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

  The present invention relates to a method for composing an image in an apparatus for storing an image such as a multimedia document.

  Data composed of a plurality of types of data, such as text data, audio data, image data, and moving image data, is generally called multimedia data. 2. Description of the Related Art Conventionally, many systems have been proposed in which a character / numerical database of information for explaining and assisting multimedia data is created and a pointer to the multimedia data is prepared as one of the data items.

  For example, in the case of image data, IMDS (see Non-Patent Document 1) and LIMS (see Non-Patent Document 2) have been developed as storage devices for a Landsat image database.

  In addition, as described in Patent Document 1, for example, one image is composed of not only one piece of image data but also a plurality of parts image data, and the content of the data is used when searching. In addition, a system has been proposed that supplements the prior art with how to assign a keyword and the lack of priority in a search.

  In addition, there is also an example of image data, in which a content-dependent data search is performed using relation information describing the mutual relation of image part data. Since GRAIN of the University of Illinois (see Non-Patent Document 3) regards an object inside each image as a minimum circumscribed rectangular area and represents it in a table of [object name-inclusion relation-position coordinate-attribute], for example, A search using the position coordinates of the logical image can be performed.

  Further, as a search method for referring to the contents, there is a method as shown in Patent Document 2. The image information search device described in the document uses, as search information, components constituting image information and sizes, orientations, postures, positional relationships, etc. in the image information instead of identification symbols or keywords. In the key image search method using various attributes described above, the user himself / herself can easily change the criterion for determining the degree of approximation for each search.

Further, when data such as a photograph and characters are mixed, all of them are treated as one data type, for example, an image. For example, as shown in Patent Document 3, compression encoding is performed and stored. There is also a method in which only a block is decoded at a high resolution, a human reads text data, and determines whether or not the file is a target file from an image and a character string written therein.
JP-A-3-139768 JP-A-2-51775 JP-A-3-87977 Y. E. FIG. Lien, et al. : "Design of an Image Database", Proc. of Picture Database Description and Management 77, p. 131 (1977) Shinoda, Kido; Image Database Science Journal, 63, 12, (1981) N. S. Chang: "Image Analysis and Image Database Management", UMI Research Press, Ann Arbar, Michigan, (1981)

  However, in these conventional methods, one specific data type among a plurality of data types such as text and image is assumed as a data type serving as a search key (for example, an image in the case of GRAIN). For example, it was difficult to automatically search for "a file in which an explanatory note of" photographed on February 14, 1988 "was additionally written in text on a chest radiograph".

  Accordingly, an object of the present invention is to make it possible to easily search target data and configure an image using a plurality of data of different data types as search keys.

  In the present invention, data to be searched is composed of a combination of a plurality of component data, and while holding component data of various data types, these are used in a complex manner, and a set of component data and component data is used. The target component data or a set of component data is retrieved from the content itself, the attribute given to them, and the description of the relationship between them, and configured as an image.

  A component data storage unit that stores a plurality of objects as component data that can be raster-drawn on an image, and an image processing operation performed when each of the plurality of objects and each of the plurality of objects is drawn on a raster image A relation storage unit for storing, for each image, a relation storage unit for storing an image, wherein the relation information of a desired image stored in the relation storage unit is stored in the image storage device. The component data required for the above configuration is retrieved from the component data storage unit, and an image processing operation defined in the related information is performed on the retrieved component data to form an image.

  According to the present invention, a desired image can be searched for and configured based on the contents of component data of a plurality of types of data. Further, as one data type is searched, the search target area is narrowed down, so that a high-speed search is possible.

  FIG. 1 is a basic block diagram for explaining the outline of the present invention.

  As shown in FIG. 1, in the multimedia document storage device of the present invention, as a storage unit, a component data storage unit 1 for storing each component data and a set of component data, and a set of each of the component data and the component data And a relation storage unit 3 for storing a description of the relationship between each component data and a set of component data. As a search unit, a content search unit 4, an attribute search unit 5, a document / relation search unit 6, and a compound search unit 7 for obtaining a desired set of component data using the three search units 4, 5, and 6 It has. The content search unit 4 includes a text search unit 4a and an image search unit 4b as described later. It further includes a processor 8 for calculation, a display device for displaying / editing search results, a keyboard for key input, a pointing device such as a mouse / digitizer, and a multimedia data input / output unit 9 for voice and images. ing.

  The multimedia data such as text data and image data input by the multimedia data input / output unit 9 are automatically or artificially given various attributes, and the contents of the data are stored in the component data storage unit 1 and the attributes. Are stored in the attribute storage unit 2. Attributes that are automatically assigned include date, resolution, registrant name, data size, image data address, data type, etc. Attributes that are artificially assigned include image name, access right, and expiration date , And treatment after the expiration date. The component data is retrieved by the content retrieval unit 4 and output from the multimedia data input / output unit 9, where editing such as combining / cutting / pasting is performed. The position information of each component data is included in the relation storage unit 3. Thereafter, a relation file describing a mutual relation between a plurality of pieces of component data generated as a result of editing and a performed operation is stored in the relation storage unit 3. The relation file also includes information on editing operations such as enlargement / reduction, color conversion, and affine conversion.

  From the contents, attributes, and relationships of the part data generated by the above procedure, a file is searched by the compound search unit 7 using the contents search unit 4, the attribute search unit 5, and the document / relation search unit 6, and the multimedia data It is displayed on the input / output unit 9.

  An embodiment of the present invention will be described with reference to the drawings. Members corresponding to those shown in FIG. 1 are denoted by the same reference numerals.

  As shown in FIG. 2, the content search unit 4 in FIG. 1 includes a text search unit 4a and an image search unit 4b. The multimedia data input / output unit 9 includes a microphone / speaker 9a for voice input / output, a scanner 9b for image input, a mouse 9c for coordinate input, and a display / input for text input and image display. It has a keyboard 9d and a network interface 9e for communication with the outside. Further, in order to hold data from the multimedia data input / output unit 9, the component data storage unit 1 stores image component data storage unit 1a, text component data storage unit 1b, and audio component data storage unit as shown in FIG. 1c.

  With the above configuration, for simplicity, first consider a multimedia data document in which image data and text data are mixed.

  First, the operation at the time of registration will be described. The image data (for example, a tree) input from the scanner 9b is first registered as one file in the image component data storage unit 1a. FIG. 4A schematically shows the image data of the registered tree. Other image data, for example, image data of people, roads, and the like are similarly registered as necessary. The image component data storage unit 1a has a capacity enough to store a plurality of image data. For example, each image data is sequentially written from a head address. Each image data is given a unique identifier, for example, a file name such as “image 1”, “image 2”,. Note that the file name itself is stored in the attribute storage unit 2 and is managed together with the head address of the corresponding image data. FIG. 4B schematically shows the registration state of the plurality of image data. In this example, the image data of the tree is registered as “image 2”. The relationship between the file name and the address where the image data is actually stored is stored in, for example, a correspondence table between the file name and the address pointer. At the time of registration, each image data is provided with a keyword indicating the content of the image data. For example, the keyword “tree” is attached to the image data of “image 2” and stored in the attribute storage unit 2.

  Next, when a keyword "tree" is input through the keyboard section of the display / keyboard 9d for editing, the keyword and the keyword of each image data registered in the attribute storage section 2 are sequentially compared. The tree registered as "image 2" is searched. By designating the display position of the tree image data with the mouse 9c, the tree image data ID2 is superimposed on the image being edited displayed on the display unit of the display / keyboard 9d (FIG. 4 (c)). )reference). The same applies to other image components, for example, image data ID3 and ID4 of people and roads. The coordinate data on the display screen of each image data is stored in the relation storage unit 3 described later as data accompanying each image data, and is retained in the memory until the editing operation is completed. As the data indicating the position of the image, for example, the coordinates of a point at the upper left corner of a rectangle circumscribing the image can be used.

  Further, the text data TD “<CR> taken on October 7, 1990” input from the keyboard part of the display / keyboard 9d is converted to the mouse 9c of the image being edited as shown in FIG. Is written in an arbitrary part designated by the above, and when the writing is completed, it is registered as "text 1" in the text component data storage unit 1b (see FIG. 4D). <CR> is a control symbol indicating a line feed. Although not shown, the same applies to other text data, and a plurality of text data are sequentially registered in the text component data storage unit 1b. The data of the coordinates of each text data on the display screen is stored as data accompanying the text data.

  As described above, the creation of one document is created by combining the image and the text, and when the end of the document creation is instructed, a file name, for example, “document 1” is given to the document. Editing work is similarly performed for other documents.

  Finally, when the editing operation is completed, a file (hereinafter referred to as a relation file) indicating which relation is used for each created document and which part data is used is generated. (See FIG. 4E). In the example of FIG. 4E, “text 1 (x1, y1) + image 2 (x2, y2) + image 3 (x3, y3) +...” Is registered as a relation file corresponding to “document 1”. Have been. In this relation file, “document 1” includes “text 1” at coordinates (x1, y1), “image 2” at coordinates (x2, y2), and “image 3” at coordinates (x3, y3). It shows that it is composed by combining. Actually, the contents of the relation file include not only the coordinates of each component data but also the enlargement / reduction ratio for describing how large each component data is with respect to other component data. The information also has the relative ratio indicating

  Next, the operation at the time of document search will be described with reference to FIG.

  For example, when an attempt is made to search for “document 1”, the relation storage unit 3 (see FIG. 5A) first searches for a relation file with the file name “document 1”, Obtain information about the relationship between data. That is, from the relation file shown in FIG. 5B, “document 1” includes “text 1” at coordinates (x1, y1), “image 2” at coordinates (x2, y2), and coordinates (x3, y3). ) Is obtained by synthesizing “image 3” and the like.

  Next, the part data necessary for document reconstruction obtained from the relation file is retrieved from the part data storage unit 1, and the multimedia document is reconstructed based on the relation between the parts in the relation file. In the example shown in the figure, "text 1", "image 2", "image 3",... Are necessary to reproduce "document 1". The data of “image 2”, “image 3”,... Is searched from the image component data storage unit 1a (see FIG. 5 (d)). A multimedia document as shown in FIG. 5E is constructed by referring to the coordinates of each data component data.

  The above is the outline of the operation of the document storage device that is the basis of the present invention. Hereinafter, the characteristic portions of the present invention will be described.

  In the above example, first, the relation storage unit 3 is referred to, and “document 1” is reconstructed based on the relation file having the file name “document 1” therein. When it is desired to search for a desired document from the contents of the text part "Shooting Date", a text search provided in the contents search part 4 of FIG. 2 is performed from the text parts data storage part 1b of the parts data storage part 1 shown in FIG. The text content is searched by the unit 4a. As a result, "text 1", which is a text component having the content "<CR> shot on October 7, 1990 in Yokohama", is obtained. Then, referring to the relation storage unit 3, a relation file of "Document 1" using "Text 1" is obtained, and "Document 1" is reconstructed based on this.

  Similarly, when a user wants to search for an image in which a tree is drawn, first, the image search unit 4b in the content search unit 4 reads the tree part data from the image part data storage unit 1a stored in the part data storage unit 1. Search for. This search can be performed using a keyword when a keyword is attached to each component data. When a keyword is added, the comparison is performed by comparing the reference bitmap pattern of the tree with the bitmap pattern of the image component data. As a result of this search, it is found that “image 2” is image part data of a tree. Next, by searching the relation storage unit 3 for a relation file using “image 2”, it is found that “document 1” uses “image 2”. Finally, “Document 1” is reconstructed from “Text 1”, “Image 2”, “Image 3”,... Constituting “Document 1”.

  In the above, the characteristic part of the present invention has been described by taking the image data and the text data as examples, but the same applies to the case of audio.

  Next, a case of performing a compound search of these will be described. Here, the compound search means a search using a plurality of data of different data types. In the following example, a compound search is performed using text data and image data as search keys.

  Now, it is assumed that the content of the related file of each document is as shown below.

Document 1: text 1 + image 2 + image 3 + ... +
Document 2: text 2 + image 1 + image 8 + ... +
・ ・ ・ ・ ・ ・ ・ ・ ・
Document 7: text 7 + image 5 + image 6 + ... +
・ ・ ・ ・ ・ ・ ・ ・ ・
Document 11: text 11 + image 8 + image 22 + ... +
Document 12: text 12 + image 7 + image 9+... +
Document 13: text 13 + image 10 + image 15 + .. +
Using the above example, when searching for "landscape with a tree-lined road taken in 1990", first, the text "1990" and "photographed" are searched from the text component data storage unit 1b. Assuming that "text 1", "text 7", "text 12", and "text 13" include text data of "1990" and "photographing", these text data are included in the related file. When a document is requested, "document 1", "document 7", "document 12", and "document 13" are obtained.

  Next, it is assumed that, when the image part data of the row of trees or the tree is searched from the image part data storage unit 1a, "image 2," "image 8," and "image 22" are applicable.

  When a document containing these texts and images in the related file is searched from the above four documents, "Document 1" is obtained.

  Further, the operation of the characteristic part of the present invention will be described with reference to a flowchart.

  As shown in FIG. 6, first, when a search request is received (step 101), it is determined whether or not the request requires compound processing (step 102). If necessary, compound search processing is performed (step 103). ), And display the search result (step 104). Details of the compound search process will be described later.

  Otherwise, which search process is requested is checked in order (steps 105 to 107), the search process is performed accordingly (steps 105 to 107), and the search result is displayed (step 104). .

  If none of the search processes is performed while receiving the search request, an error message is displayed (step 111) and the process ends.

  Next, the compound search processing will be described with reference to FIG.

  In the complex search process, whether to perform an attribute search (step 201), a relation search (step 203), or a content search (step 205) is checked in order, and if necessary, each search is executed ( Steps 202, 204, and 206) The process proceeds to the next step while narrowing down the target search area.

  Note that the attribute search and the relation search here are for searching for a desired document from the attribute attached to the part data or relation data defining an operation between the part data, and are similar to the text search described later. The full text search is performed using the “grep” command of the Unix (registered trademark) operating system. In addition, the search efficiency is improved by retaining values such as attribute values in an existing relational database. Of course things are possible.

  Examples of the attribute search include a search using an image registrant name, a search using a keyword, a search for a document to which the searcher has access right, a search using an image name, and the like. In addition, examples of the relation search include those in which the coordinates of the component data in the left corner are (10, 25), those in which a tree and a person are arranged side by side, and other component data such as an apple in a tree image. There are things that have been done.

  Further, regarding the content search, since a plurality of content searches may be performed depending on the data type, it is determined whether or not the search has been completed after the processing (step 207).

  Regarding the content search, as shown in FIG. 8, it is determined whether the component is a text component (step 301), a voice component (step 302), or an image component (step 303). If there is no search processing corresponding to the request (steps 304, 305, and 306), an error message such as "unknown parts data type" is displayed (step 307), and the process ends.

  Regarding the actual content search method such as text search / voice search / image search, any appropriate existing method may be used, but simply the command “grep” used on the Unix (registered trademark) operating system is used. "And" binary (comparison) search by pattern matching using the command "cmp". In particular, regarding image content search, as disclosed in Japanese Patent Application Laid-Open No. 2-293977, the feature amount of a plurality of registered images to be searched and the feature amount of a condition image given as a search condition are determined. If a method of calculating the degree of coincidence and determining that a search target exists when the degree of coincidence is within a predetermined allowable range is used, it is possible to perform a search using various feature amounts of an image including colors and the like.

  Although not described in the present embodiment, an input / output device, a component data storage unit, and the like can be provided for a moving image and handled in the same manner.

    In the above-described embodiment, a plurality of components are combined and registered as one document in the related file. However, a plurality of components may be combined and registered as a combined component in the related file.

  Further, a second embodiment will be described using a structured image as an example.

The purpose of the structured image itself is
1. Provided is an image format that can be operated interactively with a minimum of memory using images of a display resolution and can generate a high-resolution and high-quality output image.
2. It allows you to handle the acquisition, storage, editing / adjustment, and delivery / transmission of an entire gamut of objects to render output.
3. Reduce disk space. In particular, an image having a high reusability is a structured image, in which only the original image is stored.
4. By storing data before drawing, high-quality images are provided.
5. By saving the image operation itself, the image operation can be reused or the operation can be edited.
And so on.

  What corresponds to the component data of the present invention in the structured image is an “object” representing data that can be drawn on a raster image. The "object" is rendered on a "paste board" having a role like a canvas in a painting by performing a required image processing operation.

Here, the image processing operation is a device-independent description of an image processing instruction, and most of the description depends on the implementation. Among them, the image processing operation as a desirable basic function is as follows. There is something.
1) Transform: spatially deforms the size and shape of an image.
2) Formatter: Converts pixels from one format to another.
3) Filter: performs various filtering operations.
4) Colorizer: adjusts the color value of the image.
5) Mask: Apply a mask operation.
6) misc. : Apply other various operations.

  A structured image is a hierarchical description of a composite raster image composed of objects and image processing operations in this way.

  This description is a file called a structured image description, and the relation information stored in the relation storage unit of the present invention corresponds to this file.

  The structured image does not have an absolute size in the structured image description, and the size of each child object is stored in the pasteboard as a relative size to the parent of the sub-object.

  In order to convert the structured image object into a raster image, for example, the drawing process is started by an appropriate control command using the absolute size of the parent object or the like as a variable from the display / keyboard device 9d together with the object and the pasteboard. The rendered image object and all child objects are automatically drawn to the appropriate output size.

  Any data representation that can be drawn on a raster image can be a structured image object (see FIG. 9A). This data is negotiated between Aldas Corporation and Microsoft, and is a simple raster such as a Tag Image File Format (TiFF) file, which is a tag-based image format widely used for image exchange between personal computers. It may represent an image, or it may represent more complex data composed of a hierarchical set of child objects and associated raster processing operations.

Object data types that can be handled include:
1) Raster image-TiFF or other digital display, etc. 2) Toy text-Simple text description 3) CGM (Computer Graphics-Metafile for the Storage and Transfer of Picture Description Graphics Simple-graphic representation) ) IPD (Image Processing Definition)-Structured Image Definition 5) MR (Multiple Representation) File-There is a file including a description of a multiple image expression.

  Among them, the raster image is stored in the image part data storage unit as shown in FIG. 3A, the toy text is stored in the text part data storage unit in FIG. 3B, and the other data types are also stored in the text part data storage unit. Each is stored in the component data storage unit supporting the data type.

  Here, to supplement the MR file a little, one structured image object is represented by one or more data types and data files (see FIG. 9B). For example, one structured image object may refer to a directory containing a plurality of TiFF files stored at different resolutions. Each TiFF file represents the same image, but has different absolute pixel sizes and brightness. These concepts are known as multiple representations. In the structured image, multiple expressions are treated as dynamically determined. Note that “sunras” in the figure means an image format type (Sun Raster file format) used by Sun Micro Corporation.

  In the procedure for actually drawing a structured image object as a raster image on such a pasteboard, first, the pasteboard, the structured image object, and the absolute size of the parent object are passed as variables.

Several operations are performed to render an appropriate raster image. That is,
1) Determination of the absolute size of the pasteboard 2) Determination of all expressions referenced by the object 3) Selection of an appropriate expression to draw based on available information 4) Fit mode, justification, etc. Activate the appropriate drawing procedure according to the above, and draw the data on the raster. 5) Rotate the raster drawn at the angle specified by the Angle attribute. 6) Adjust the position of the control point.

  Then, the absolute size of the pasteboard is determined by multiplying the size of the parent by the relative size specified in the pasteboard.

  The structured image object given here actually exists as a reference to a data file or a representation that can be drawn on a raster image (such as text). Also, one object can refer to multiple expressions, each representing the same image, but may have different storage formats, sizes, or resolutions. A search is made to determine which expression is the "optimal" expression to render. The optimal expression here means, for example, an image file that can be drawn at the fastest speed because the image enlargement / reduction ratio differs depending on the size of the image file among a plurality of image files stored at different resolutions. The expression by the choice.

  When the representation is retrieved, it is rendered into a raster image, but must have a rendering procedure to convert it to a raster image for each data type of the supported representation.

When locating the object to be drawn on the output raster image and returning the data, six operations are required along with the drawing of the structured image. That is,
1) Determine the position of all expressions of the object 2) Specify the type of each expression such as TiFF, text, graphics 3) Determine the attributes of each expression necessary to select the optimal expression 4) Use each Applying a value function to a possible expression and selecting an expression with a high value 5) Return a pointer to the expression data 6) Return the type of the expression data.

In the current syntax of the structured image description language, the structured image object is one of the following three types.
(1) Object = ｛
External = ｛
System = "AAA"
Name = “bird.TiF”;
｝;
｝;
(2) Object = ｛
Internal = ｛
Bytecount = 1024;
Data = ｛
. . .
｝;
｝;
｝;
(3) Object = ｛
IPD = ｛
AspectRatio = 1.5;
DefaultWidth = 4in;
. . .
｝;
｝;

  An object defined as External is data referenced by the object and is not stored in the current structured image description. Generally, the actual data is stored as a separate file. The system field is a description used to know how to locate the data.

  A name entry is a set of keywords used to actually locate the requested data. In the above example, the system entry "AAA" indicates that the actual data can be found by searching for the path on the appropriate system AAA file system for the file whose name entry is "bird.TiF". Is implemented to be interpreted as: If the system entry is "BBB", the keyword in the name field is handled as a keyword when searching data in the BBB database.

  An object defined as Internal contains data stored directly in the structured image description. This is useful for storing small objects, such as raster icons and simple text files, directly in the structured image description and does not require external storage.

  A structured image definition is one of the object data types of a structured image that allows the structured image to consist of other structured image objects and image processing operations.

  Based on the information stored in the structured image object, all available representations must be located and the data type and attributes required to select the "optimal" representation for rendering must be determined. Must. In the present invention, these can be determined by a search request such as inquiring directly to the attribute search unit using a keyword in the name field.

  Once all expressions have been determined, the most appropriate expression to render on the output raster image is selected. This mechanism is implemented by applying a value function computation to each expression. The exact value function is implementation dependent and is not described in the structured image. However, it should be based on the attributes of the representation, the attributes of the output image required, the target output device (ie monochrome printer or color display, etc.) and the complexity of the processing. Once the representation is chosen, it returns a pointer to the actual data and the data type so that processing can begin.

  To explain the structured image definition a little more, the structured image definition represents a reference to element objects and a structured description of an image processing operation applied to the element objects. FIG. 10 is a graphical representation of a structured image definition.

The structured image definition consists of the following descriptors:
1) Default size 2) ReadObject
3) Pass 4) Image processing operation 5) Merge 6) Paste board 7) Initializer
And 8) Component pasteboard.

  In FIG. 10, each vertical line is called a path and is a group of image processing operations. ReadObject represents a procedure that converts the structured image object into a raster image suitable for processing by the image processing operations in the pass. The drawn object is processed in sequence for each image processing operation through a path. Each path stores therein a pasteboard used by ReadObject to define a “frame” in which the structured image object is drawn. This pasteboard is referred to as a component pasteboard. The horizontal line is the structured image definition object pasteboard, representing the "empty" raster image into which the component pasteboard is merged. The black triangle represents a merge operation that indicates where the component pasteboard is bound to the object pasteboard. The order in which the component pasteboards are merged is clear, and therefore images with meanings such as "up" and "down" are supported. FIG. 11 illustrates the flow of raster data during a structured image (that is, the flow of control from ReadObject to merge).

  Now, when the structured image as described above is handled by the multimedia document storage device as shown in the present invention, what is stored in each storage unit of the block diagram shown in FIG. 1 will be described.

  Now, consider a case where a structured image generated by another system is transferred to the multimedia document storage device in the form of a structured image description language through a network interface (9e in FIG. 2).

  The relationship storage unit stores the structured image definition as a result of the interpretation of the structured image description, and the attribute storage unit automatically stores the storage date and time, registrant name, data size, etc. A keyword for searching for an object described in the name field in the structured image description is recorded. These are used for later search. In the component data storage unit, a structured image object defined as External is selected and its contents are stored. At the same time, the name of the multimedia document storage device is described in the system field defined in the External of the structured image description instead of AAA or BBB originally described.

  Also, for the object defined as Internal, the content itself of the object described in the data field is copied and stored in the component data storage unit for later retrieval.

  At the time of retrieval, according to the flowchart of FIG. 6, first, a retrieval request is accepted (step 101), and it is determined whether or not it requires compound processing (step 102). For example, an image A and a text B are drawn. When a complex search is required, as in the case of a structured image, a complex search process is performed (step 103), and the search result is displayed (step 104).

  Otherwise, which search process is requested is checked in order (steps 105 to 107), the search process is performed accordingly (steps 108 to 110), and the search result is displayed (step 104). .

  If none of the search processes is performed while receiving the search request, an error message is displayed (step 111) and the process ends.

  Next, the compound search processing will be described with reference to FIG.

  In the complex search process, whether to perform an attribute search (step 201), a relation search (step 203), or a content search (step 205) is checked in order, and if necessary, each search is executed ( Steps 202, 204, and 206) The process proceeds to the next step while narrowing down the target search area.

  As described in the first embodiment, the attribute search here is to search for a desired document from the attribute added to the component data. In the relation search, the document is stored in the relation storage unit. An image is reconstructed from the obtained structured image definition by referring to the structured image definition and searching for a desired structured image definition.

  In the relation search process 204, a desired image structure such as an image processing operation and a merge point of each object as described in the description of the structured image is separately edited by an editor (this is a Unix (registered trademark) operating system). A structured image description may be created with a “vi” editor, or a dedicated structured image editor may be prepared) and compared using the “grep” command of the Unix (registered trademark) operating system. Alternatively, search efficiency may be improved by storing the information in a relational database or the like in advance.

  Further, regarding the content search, since a plurality of content searches may be performed depending on the data type, it is determined whether or not the search has been completed after the processing (step 207).

  Regarding the content search, as shown in FIG. 8, it is determined whether the component is a text component (step 301), a voice component (step 302), or an image component (step 303). If there is no search processing corresponding to the request (steps 304, 305, and 306), an error message such as "unknown parts data type" is displayed (step 307), and the process ends.

  As with the first embodiment, any conventional technique may be used for the content search. For example, for example, in the content search of an image part, the same raster image as the image A to be searched from the scanner 9b is used for the content search of the image parts. Scan input is performed, and the contents are sequentially compared with data stored in the image part data storage unit. If image part data having desired contents is found in the image search unit 4b, the structure referring to the image part data is searched. The structured image is reconstructed from the structured image description stored in the relation search unit, and the result is displayed.

FIG. 1 is a basic block diagram for describing an outline of the present invention. FIG. 2 is a block diagram of an embodiment of the multimedia document storage device of the present invention. FIG. 3 is a schematic diagram illustrating a detailed configuration of a component data storage unit. FIG. 4 is an explanatory diagram showing a basic concept at the time of registration in the present invention. FIG. 4 is an explanatory diagram showing a basic concept at the time of search in the present invention. 5 is a basic flowchart at the time of retrieval according to the present invention. It is a flowchart of a compound search process. It is a flowchart of a content search process. 5 is a graphical representation showing the various data types that make up a structured image object. 7 is a graphic representation of an image processing expression illustrating the relationship between references to configuration objects and output structured images representing a structured description of image processing operations applied to those configuration objects. FIG. 4 is a diagram illustrating a flow of raster data during drawing of a structured image.

Explanation of reference numerals

  DESCRIPTION OF SYMBOLS 1 ... Part data storage part, 1a ... Image part data storage part, 1b ... Text part data storage part, 1c ... Sound part data storage part, 2 ... Attribute storage part, 3 ... Relationship storage part, 4 ... Content search part, 4a ... Text search unit, 4b ... Image search unit, 5 ... Attribute search unit, 6 ... Document / relation search unit, 7 ... Compound search unit, 8 ... Processor, 9 ... Multimedia data input / output unit, 9a ... Microphone / speaker, 9b: scanner, 9c: mouse, 9e: network interface

Claims (1)

A component data storage unit that stores a plurality of objects as component data that can be raster-drawn on an image, and an image processing operation performed when each of the plurality of objects and each of the plurality of objects is drawn on a raster image A relation storage unit for storing the obtained relation information for each image, a method of configuring an image in an image storage device,
From the relation information of the desired image stored in the relation storage section, component data necessary for the configuration of the image is retrieved from the component data storage section, and the image data defined in the retrieved component data is stored in the retrieved component data. An image forming method in an image storage device, wherein an image is formed by performing a processing operation.

Images