JPH11232301A - Method and device for retrieval canceling difference of image data base and recording medium recording image data base difference canceling retrieval program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To attain retrieval canceling a difference between the expression formats of image features or a difference between the transmitting/receiving methods of image data by converting an inquiry sentence from a user into an expression format suited to each image data base, translating the sentence of the converted expression format and retrieving the image data base based on the translated sentence. SOLUTION: A conversion function for converting between domains is prepared (S15) and a user domain is determined (S17). Then processing goes to a retrieving phase, an inquiry sentence from a user is analyzed and a column table is retrieved to determine an inquiry data base (S21). The domain of a column or a virtual column included in the inquiry sentence is retrieved from the column table (S23), the user domain is retrieved from the inquiring user and its domain (S25), a conversion function from the user domain to a local domain is searched (S27), the inquiry sentence is translated to an inquiry sentence to an image data base by using the conversion function (S29), and the data base is retrieved by the translated inquiry sentence.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION The present invention relates to an image processing system in which a plurality of image databases have the same image characteristics (hue, outer shape, saturation, brightness,
It is possible to specify textures, layouts, etc., but the expression format of image features that can be specified as a search condition differs for each image database, or a method of transferring image data as a search result to an application program for each image database (file transfer) , Parameter passing, etc.), an image database heterogeneity resolving search method and apparatus for realizing a search that eliminates a difference in expression form and a difference in image data delivery method, and a recording in which an image database heterogeneity resolving search program is recorded Regarding the medium.

[0002]

2. Description of the Related Art In recent years, the performance and price of personal computers have been reduced,
With the widespread use of open networks such as the Internet, intranets, and extranets, an environment in which image data can be easily viewed and processed by end users has been rapidly established. At the same time, in order to efficiently manage a large amount of stored image data, centrally manage the image data and specify image features such as hue, outer shape, saturation, brightness, texture, layout, etc. Research and development for realizing a search for more similar image data have become active. In fact, the top vendors in relational database shares have shifted their targets to object relational databases and have begun to provide libraries as cartridges that can easily handle image data.

[0003] However, there is no standardized interface for providing a search for similar image data by designating image characteristics in an image database, and each vendor provides its own interface. Is the actual situation. In such a situation, in an environment where a plurality of image databases exist, in order to search for image data that is uniformly similar from the plurality of image databases by specifying image characteristics, it is necessary to use an application program for each system. was there. However, in this method, the user needs to write a program unique to the system, and the expression form of the image feature that can be specified in the search condition and the method of delivering the image data of the search result are fixed for each system. .

As another means, there has been a method using a multi-database technology which is a technology for integrating relational databases. Multi-database technology is a technology that allows users to search distributed databases without having to be aware of the expression format or structure of individual information by predefining the virtual structure of the database used by the user, called the global schema. It is. However, in this method, the user can view data only in the form of the global schema, and is forced to take one view.

[0005]

In the conventional method, in order to create an application program or define a global schema, an image feature expression format that can be specified as a search condition of an entire image database and delivery of image data of a search result. Needed to know how. However, it requires access to the contents of all image databases, which is actually very difficult. Therefore, even if you do not know as much as possible the expression format of image features that can be specified as search conditions other than your own image database and the delivery method of image data, you can specify the expression format and search format of image features that can be specified as search conditions for each image database. There is a need for a mechanism that can eliminate the difference in the delivery method of the resulting image data.

[0006] Further, the user can express an image feature in an expression form (for example, term, image feature value,
There is a request to search the information of the entire image database by specifying a file or the like, or a request to obtain image data of a search result of the entire image database by an individual delivery method (file transfer, parameter transfer, etc.).

In this method, the user wants to use the data of another image database by using the application program used by the user. However, the application program can use the expression form or the search result of the image feature which can be specified in the search condition of the image database. This is because it is made in accordance with the image data delivery method. If this request is to be implemented using existing multi-database technology, a global schema is defined for each user that uses the same image feature expression format and search result image data delivery method that can be specified in the user's image database and search conditions. And there is a problem of inefficiency.

Therefore, without defining a global schema, it is possible to eliminate the difference by defining the expression form of image features that can be specified fragmentarily in search conditions and the method of delivering image data of search results. is necessary. By providing a mechanism that allows the user to easily change the expression format of image features that can be specified in the search conditions that the user wants to use and the delivery method of the image data of the search result, the difference in the expression format and the delivery method of the image data are provided. There is a demand for a search that eliminates the differences.

[0009] The present invention has been made in view of the above,
Its purpose is to record an image database dissimilarity search method and apparatus and an image database dissimilarity search program that realizes a search that eliminates differences in the expression form of image features and differences in the method of transferring image data. To provide a medium.

[0010]

To achieve the above object, according to the present invention, in an environment where a plurality of image databases exist, each image database can specify the same image feature as a search condition. When the expression format of image features that can be specified as search conditions differs for each image database, or when the method of delivering image data that is a search result to an application program differs for each image database, the difference in expression format and delivery of image data This is an image database dissimilarity resolving search method that realizes a search that eliminates the difference between the methods, wherein a column that exists in each image database and a column that can be specified as a search condition but the corresponding column does not exist in the image database Data that can be specified in each image database search request Local domain management table for each image database, with the expression format of the eyes, the expression format of image features that can be specified as search conditions for each image database, and the domain that is the method of delivering image data that is the search result of each image database as the local domain Each domain with the same expression format and delivery method is managed as a domain group in the domain group management table, and the conversion function for converting between domains is managed in the conversion function management table and specified by the user. The expression format of the data used in the search conditions, the expression format of the search result data requested by the user, and the domain that is the method of delivering the search result data requested by the user are managed as user domains in the user domain management table for each user. Inquiry entered by the user An image database in which sentence search data is stored and a local domain corresponding to the search data are searched from the local domain management table, and a user domain is managed from the user name and a domain group of the search data designated by the user. Search from the table and if the searched local domain and user domain are different,
A conversion function for converting the user domain to the local domain and a conversion function for converting the local domain to the user domain are searched from the conversion function management table, and the query sentence is converted into each image database using the searched conversion functions. The purpose of the present invention is to convert the data into an expression format that conforms to the above, translate it into a query sentence for the image database, and search each image database using the translated query sentence.

According to the first aspect of the present invention, an image database storing search data of an inquiry sentence input by a user and a local domain corresponding to the search data are searched from a local domain management table, A conversion function for searching a user domain from a user domain management table for a user domain from a user name and a domain group of search data designated by the user, and if the searched local domain is different from the user domain, a conversion function for converting the user domain to the local domain; A conversion function for converting a local domain into a user domain is searched from a conversion function management table, and a query sent from a user is converted into an expression format suitable for each image database using the former conversion function, and an inquiry to the image database is made. Translate it into a sentence Searching each image database using the translated query, and using the latter conversion function to convert the expression form and delivery method of the search results to the user domain, the expression of image features specified in the search conditions of each image database Even if the delivery method of the image data serving as the search result to the application program for each format or image database is different from the expression format or delivery method specified by the user, the search can be freely performed.

According to a second aspect of the present invention, in an environment where a plurality of image databases exist, each image database can specify the same image feature as a search condition, but each image database can specify the same image feature as a search condition. If the expression format of the features is different or the delivery method of the image data that is the search result for each image database to the application program is different, the image that realizes the search that eliminates the difference in the expression format and the difference in the delivery method of the image data A database dissimilarity resolving search apparatus, which manages a virtual column in which an existing column of each image database and a column that can be specified as a search condition but a corresponding column does not exist in the image database, Representation format of data items that can be specified in database search request and each image database A local domain management table that manages, for each image database, a domain as a local domain, which is an expression format of image features that can be specified as search conditions and a method of delivering image data that is a search result of each image database, and an expression format and delivery method A domain group management table that manages each domain having the same target as a domain group, a conversion function management table that manages a conversion function that performs conversion between domains, an expression format of data used in a search condition specified by a user, A user domain management table that manages, for each user, a domain as a user domain, which is a representation format of search result data requested by the user and a method of delivering search result data requested by the user, and a query sent from the user. Search data is stored Domain search means for searching the image database and the local domain corresponding to the search data from the local domain management table, and searching the user domain management table for a user domain from a user name and a domain group of the search data specified by the user; and When the local domain and the user domain searched by the domain searching means are different, a conversion function for converting the user domain to the local domain and a conversion function for converting the local domain to the user domain are searched from the conversion function management table. Using the conversion function search means and the conversion function searched by the conversion function search means, the query sentence is converted into an expression format suitable for each image database, and translated into a query sentence for the image database. Question The gist of the present invention is to have a search means for searching each image database by the combined sentence.

According to the present invention, an image database storing search data of a query sentence input by a user and a local domain corresponding to the search data are searched from a local domain management table, A conversion function for searching a user domain from a user domain management table for a user domain from a user name and a domain group of search data designated by the user, and if the searched local domain is different from the user domain, a conversion function for converting the user domain to the local domain; A conversion function for converting a local domain into a user domain is searched from a conversion function management table, and a query sent from a user is converted into an expression format suitable for each image database using the former conversion function, and an inquiry to the image database is made. Translate it into a sentence Searching each image database using the translated query, and using the latter conversion function to convert the expression form and delivery method of the search results to the user domain, the expression of image features specified in the search conditions of each image database Even if the delivery method of the image data serving as the search result to the application program for each format or image database is different from the expression format or delivery method specified by the user, the search can be freely performed.

Further, according to the present invention, in an environment where a plurality of image databases exist, each image database can specify the same image feature as a search condition, but each image database can specify the same image feature as a search condition. If the expression format of the features is different or the delivery method of the image data that is the search result for each image database to the application program is different, the image that realizes the search that eliminates the difference in the expression format and the difference in the delivery method of the image data Manages a virtual column that is a recording medium that stores a database dissimilarity resolving search program, and that is a column that can be specified as a search condition and an existing column of each image database, but the corresponding column does not exist in the image database. And data items that can be specified in search requests for each image database The domain, which is the expression form of image features that can be specified as search conditions for each image database and the method of delivering image data as search results for each image database, is managed as a local domain in the local domain management table for each image database. , Each domain having the same expression format and delivery method is managed as a domain group in the domain group management table, conversion functions for converting between domains are managed in the conversion function management table, and the search conditions specified by the user are used. The user uses a user domain management table for each user as a user domain, which is a representation format of data to be used, a representation format of search result data requested by the user, and a method of delivery of search result data requested by the user. Search for the query sent from And a local domain corresponding to the search data is searched from the local domain management table, and a user domain from the user name and a domain group of the search data designated by the user is searched from the user domain management table. When the searched local domain is different from the user domain, a conversion function for converting the user domain to the local domain and a conversion function for converting the local domain to the user domain are searched from the conversion function management table. Using the searched conversion function, the query sentence is converted into an expression format suitable for each image database, translated into a query sentence for the image database, and image data for searching each image database with the translated query sentence Be The gist of the present invention is to record a search program for eliminating heterogeneity on a recording medium.

According to the third aspect of the present invention, an image database in which search data of an inquiry sentence input by a user is recorded and a local domain corresponding to the search data are searched from a local domain management table. A conversion function for searching a user domain from a user domain management table for a user domain from a user name and a domain group of search data designated by the user, and if the searched local domain is different from the user domain, a conversion function for converting the user domain to the local domain; A conversion function for converting a local domain into a user domain is searched from a conversion function management table, and a query sent from a user is converted into an expression format suitable for each image database using the former conversion function, and an inquiry to the image database is made. Translate it into a sentence Each image database is searched by the translated query sentence, and an image database dissimilarity search program for converting the expression form and delivery method of the search result into a user domain using the latter conversion function is recorded on a recording medium. Therefore, the distribution property can be improved by using the recording medium.

[0016]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an image database dissimilarity search apparatus according to an embodiment of the present invention. The image database dissimilarity search apparatus of the present embodiment specifies the same image feature (hue, outline, saturation, brightness, texture, layout, etc.) as a search condition in an environment in which a plurality of image databases exist. Is possible, but the expression form of the image feature that can be specified as a search condition differs for each image database,
When the delivery method (file delivery, parameter delivery, etc.) of the image data that is the search result to the application program differs for each image database, the search is performed by eliminating the difference in the expression format and the difference in the image data delivery method. Yes, manages existing columns of each image database and virtual columns that are items that can be specified as search conditions but corresponding columns do not exist in the image database, and can be specified in each image database search request It manages the expression format of various data items, the expression format of image features that can be specified as a search condition of each image database, and the delivery method of image data that is a search result of each image database (this is called a domain). Each domain for which the delivery method is the same is a domain group,
From this domain group, a global domain that is typically used as a representative is selected, and a data expression format used in a search condition specified by the user, a data expression format of a search result requested by the user, or a user request. 1 in each domain group that is the method of delivery of search result data
If each domain is a user domain, and the local domain and the user domain of the image database are different at the time of search, a conversion function between the domains is called, and a query sentence of the user is adapted to each image database using this conversion function. The image database is translated into a query sentence for the image database, the image database is searched by the translated query, and the search result is converted into a user domain. As shown in FIG. 1, a syntax analyzer 1 for analyzing the syntax of a query sentence 13 input from a user's application program 12, a query processing controller 2 for controlling the overall flow of processing for this query, and a conversion of a query. Conversion unit 3 that performs A conversion function library 4 which is a group of functions for performing data conversion of the above, an inquiry processing unit 11 comprising a data processing unit 6 for converting a domain of searched data, and a database access unit 5 for accessing a distributed image database. And a metadata storage unit 7 for storing and managing information such as domains and domain groups, and information of tables included in a database.
Input of metadata to the metadata storage unit 7,
The database access unit 5 is connected to a plurality of database systems 15 each including an image database 9 and a database management system 10 for managing the image database 9. I have. Note that the image database 9 is also called a local database.

The image database heterogeneity resolving and retrieving apparatus of the present embodiment configured as described above has a schematic operation as shown in a flowchart of FIG. There are roughly two stages of processing, a preparation phase to be performed and a search phase to perform an actual search. Before giving a detailed description of the present embodiment, first, a schematic processing flow of the present embodiment will be described with reference to a flowchart shown in FIG.

In FIG. 9, first, in a preparation phase, a local domain is defined (step S11) and a domain group is defined (step S13), and a conversion function for converting between domains is created (step S15). Is determined (step S17). Then, the process proceeds to a search phase, in which a query sentence from the user is analyzed, and a column table is searched to determine a query database (step S21). Further, a domain of a column or a virtual column included in the query is searched from the column table (step S23), a user domain is searched from the user and the domain of the query (step S25), and a conversion function from the user domain to the local domain is performed. Is searched (step S27), translated into a query sentence for the image database using this conversion function (step S29), and the database is searched using the translated query sentence (step S31).

Next, each processing and the whole processing will be described in detail. First, the preparation phase will be described.
In this preparation phase, domains are defined. The concept of domain management in the present embodiment will be described with reference to FIG. As a domain, which is an expression format of data items that can be specified in a search request of each image database, an expression format of image features that can be specified as a search condition of each image database, and a delivery method of image data that is a search result of each image database, For example, as shown in FIG. 2, the domain for the hue can be expressed in various ways, such as when expressed by words (red, blue, yellow, etc.), when expressed by a 16-dimensional HUE value, or when expressed by a symbolic image file. There is.

As described above, one of various expression methods which targets the same image feature is defined as a domain group.
By setting only one of the domains as the domain used by the user, a single perspective can be provided to the user. Further, as a domain as a method of delivering image data as a search result, there are various delivery methods such as a case of delivering by a file and a case of delivering by a parameter. In this way, the various delivery methods that target the same one are defined as a domain group, and only one domain is set as a domain used by the user, so that a single A simple delivery method can be provided.

In FIG. 2, a domain group also includes an outer shape, in which various domains such as a language expression, a 24-dimensional vector value expression, and a symbolic file expression are shown. In FIG. 2, the user user1 uses a method in which the hue and the outer shape are expressed in words, and the image data uses a method of transferring the image data in a file. These are user domains of the user1.

With respect to the domain, the following four settings are required: setting of a local domain, setting of a domain group, creation of a conversion function, and setting of a user domain. To set information such as domain. These pieces of information are stored in the metadata storage unit 7. Further, the configuration of a table of each information stored in the metadata storage unit 7 is shown in FIG.
And FIG. The setting of each domain will be described.

(1) Setting of local domain: For each column existing in the local database 9 and a virtual column which is an item which can be specified as a search condition but whose corresponding column does not exist in the image database, Set the expression form of data items that can be specified in the database search request, the expression form of image features that can be specified as search conditions for each image database, and the "domain" that is the method of transferring image data that is the search result of each image database I do. This is called a local domain. For example, the domain of an image feature “hue” that can be specified by a search request of a certain image database is set as ColorFileName. This Co
The meaning of the domain lorFileName represents not only the location of the file representing hue, but also the finer form of expression of "hue". For example, this ColorFileName
Contains the host name.・ It is an absolute path. It is an expression form. The definition of a domain for a column or virtual column is held in a “column” table as shown in FIG. Basically, this is done by the local database administrator. At this time, the already defined domain is referenced using the metadata management unit 8. if,
If a domain that can be used as a domain for your database is already defined, use it and do not create a new domain.

(2) Setting of domain group: A group of each domain which is different in expression form and delivery method but has the same object is called a domain group. In this apparatus, all the domains included in this domain group appear to be unified into one domain from the user. Local database administrators determine which domain groups their defined domains fall into,
Perform the process of joining the domain group. If a suitable domain group cannot be found, create a new one. If the domain is already defined, you do not need to set it because it is included in any domain group. FIG. 3B shows a domain group. In the domain group, a domain that is normally used as one representative is set (this setting is basically performed by the administrator of the apparatus). The domain is called a global domain. FIG. 3C shows the global domain.

(3) Setting of user domain: A domain specified by a user at the time of search is selected for each user from a domain group. This is called a user domain.
When a user makes an inquiry to the present apparatus, even if there are a large number of domains, the expression format and delivery method for all user domains are unified. Further, by changing the user domain, it is possible to change how the data is viewed and handled by the user. FIG. 3D shows a user domain.

(4) Creation of a conversion function: This is not necessary if a conversion function has already been defined. A conversion function between a user domain and a local domain is created using a general programming language. Compile your function into a form that can be dynamically linked. The relationship between the conversion function and the domain is registered using the metadata management unit 8. The correspondence between the conversion function and the domain and the location of the conversion function are held in a “conversion function” table as shown in FIG.

The conversion function creates a bidirectional function such as user domain → local domain, local domain → user domain. However, depending on the conversion, it may be impossible or unnecessary to create functions in both directions. For example, in the name conversion, the conversion “Taro Yamada” → “Taro Yamada” = blank deletion is possible, but the reverse is generally impossible. Also, for a domain of an item that does not exist on the database, it can be specified as a search condition, but cannot be specified as a search item (that is,
"SELECT image WHERE hue = red" is possible,
Since “SELECT hue WHERE outline = circle” is generally not possible), only the conversion up to the query sentence need be performed, so the reverse conversion function is unnecessary. In these cases, a function in only one direction is created. The reason for creating a bidirectional function is to perform conversion by combining functions when there is no function to directly convert.

Next, the search phase will be described. A search is performed using the domain information as follows. The user generally calls and uses the functions of the apparatus from an application program in the form of a library or the like. The inquiry from the user is created in an SQL-like language which is a search language of the relational database.

(1) Determination of query database: The query is analyzed by the syntax analysis unit 1, and a database including tables, columns, and virtual columns included in the query is searched for in the “column” table of the metadata storage unit 7. know.

(2) Retrieval of a domain of a column or a virtual column included in an inquiry: A domain of a column or a virtual column included in an inquiry is found by searching a “column” table shown in FIG.

(3) Retrieval of user domain: By retrieving the "domain" table shown in FIG. 3B from the domain of (2), a domain group is retrieved, and the domain name is retrieved from the inquired user name and the domain group. , FIG.
The user domain is searched by searching the "domain group user" table shown in (d).

(4) Search for conversion function: The conversion function between the user domain and the domain of (2) is checked using the "conversion function" table shown in FIG. If a conversion function that can be directly applied to A → B is not found, a search is made to see if conversion is possible in combination with A → C, C → B. This is a search problem on a graph in which a domain is a node and the existence of a transformation function between domains is an edge, and can be easily realized using an algorithm studied in graph theory.

(5) Inquiry translation and execution: It is necessary to convert data as described below and implement a database access unit for each database management system using search items, search conditions, and the like as parameters. -Conversion of the value included in the query: The value included in the query is converted from the user domain to the local domain. -Access using an application interface for each database management system: Some database management systems support SQL-like syntax, some only provide functions, and so on. Database management using search items, search conditions, etc. as parameters Perform database access for each system.

Access is made to a plurality of image databases, and the result is the sum of the search results for each image database. Next, a specific example will be described with reference to FIGS. FIG. 5 shows a sample image database (DB) to be used. For metadata such as domains, the values in FIGS. 3 and 4 follow this example. The two databases (DB) are in the following format.

[0036]

DB1: table “painting” column of “painting” = {existent column (category, title, image), virtual column (hue, outline)} expression method of “image” = (// hostname / DB1 / picture /pict01.gif) Expression of "hue" = (//hostname/DB1/picture/color/red.gif) Expression of "outline" = (//hostname/DB1/picture/shape/circle.gif) DB2 : Table "Photo" Column of "Photo" = {Existing column (category, title, image), virtual column (hue, outline)} Expression of "image data" = (binary data) Expression of "hue" = ( 0.001667 0.000000 0.001667 0.010000 0.949167 0.037500 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000) Expression of “Outer shape” = (0.674243 0.621274 0.556349 0.469136 0.184281 0.429924 0.402778 0.504743 0.283333 0.351587 0.359079 0.755682 0.755682 0.737805 0.726985 0.624603 0.446476 0.49684 0.368 483

DB1 is a database management system that enables an image search similar to a specified file.
Is a database management system that enables an image search close to a specified value. First, as described above, the definition of the local domain, the domain group, the conversion function, and the determination of the user domain are performed as described below, respectively.

(1) Definition of local domain: DB1,
The administrator of DB2 sets the local domain for each DB column. The domain of the image in DB1 is the host name and the file name of the image.
eFileName. The hue domain of DB1 is the location of a file symbolizing the hue, and is referred to as ColorFileName. The domain of the outer shape of DB1 is the location of a file symbolizing the outer shape, and is defined as ShapeFileName. The domain of the image in DB2 is the actual data of the image, and is assumed to be PictureData. The hue domain of DB2 is a 16-dimensional HUE value representing the hue, and is referred to as ColorHUEValue. The domain of the outer shape of DB2 is a 24-dimensional vector value representing the outer shape, and is defined as ShapeVectorValue.

(2) Definition of domain group: Here, as a domain group: Picture (the domain included is PictureFileName, Pi)
ctureData, the global domain is PictureFileName
), Color (The domains included are ColorName and ColorFil.)
eName, ColorHUEValue, global domain is ColorN
ame), Shape (The domains included are ShapeName, ShapeFil
eName, ShapeVectorValue, global domain is Shap
eName) is set. Where ColorName is (red,
The hue of blue, yellow, etc.) is a domain that expresses the hue in words, and the ShapeName is the domain in which the outer shape of (circle, triangle, square, etc.) is expressed in words.

(3) Definition of conversion function: The following conversion function is defined for each domain group. Domain group: Picture Conversion function: PictureFileName2PictureData () Acquires the image data entity from the image file name. Conversion function: PictureData2PictureFileName () Stores the image data entity in a file and obtains the image file name. Domain group: Color Conversion function: ColorName2ColorFileName () Converts a hue expressed in words into a file name that symbolizes the hue. Conversion function: ColorName2ColorHUEValue () A 16-dimensional HUE indicating the hue expressed in words
Convert to a value. Domain group: Shape Conversion function: ShapeName2ShapeFileName () Converts a shape expressed in words to a file name that symbolizes the shape. Conversion function: ShapeName2ShapeVectorValue () Converts a contour expressed in words into a 24-dimensional vector value indicating the contour.

(4) Determination of user domain: In this example, it is assumed that the user domain is equal to the global domain (this is a case where the user user1 in FIG. 3D uses it). That is, (Picture user domain) = PictureFileName (User domain of Color) = Color Name (User domain of Shape) = Shape Name.

The search phase will be described. The user's inquiry is converted into an SQL-like sentence as follows. Query: SELECT image, title WHERE Hue = '
Red 'AND outline =' circle 'Meaning: Searches for an image whose hue is red and whose outline is similar to a circle and its title.

(1) Search for a table including columns of image, title, hue, and outline:
Discover DB1 and DB2.

(2) Image, title, hue, outline domain check: DB1: image → PictureFileName, title → none, hue → ColorFileName, outline → ShapeFileName DB2: image → PictureData, title → none, hue →
ColorHUEValue, outline → ShapeVectorValue

(3) Search for user domains in a domain group including each domain: DB1, DB2: image → PictureFileName, hue → Colo
rName, outline → ShapeName

(4) Search for conversion function: For (value included in conditional expression), search for (user domain → local domain conversion function).

(5) Value conversion: (value included in conditional expression)
, Execute the searched conversion function. ColorName2
ColorFileName ('red') results in '// hostname / DB1
/picture/color/red.gif '. ShapeName2ShapeFil
The result of eName ('circle') is' // hostname / DB1 / picture
/shape/circle.gif '. ColorName2ColorHUEValu
e ('red') results in '1.000000 0.000000 0.000000
0.000000 0.000000 0.000000 0.000000 0.000000 0.00
0000 0.000000 0.000000 0.000000 0.000000 0.000000
0.000000 0.000000 '. ShapeName2ShapeVectorV
The result of alue ('circle') is' 0.055137 0.013605 0.0333
33 0.064517 0.041006 0.055137 0.055137 0.013605 0.
033333 0.064517 0.041006 0.055137 0.055137 0.04100
6 0.064517 0.033333 0.013605 0.055137 0.055137 0.0
41006 0.064517 0.033333 0.013605 0.055137 '.

(6) Translation of inquiry: The inquiry sentence specified by the user is translated.

[Expression 2] DB1: SELECT image, title WHERE FROM painting WHERE hue = '// hostname / DB1 / picture / color / red.gif' AND outline = '// hostname / DB1 / picture / shape / circle.gif' DB2 : SELECT image, title FROM photograph WHERE hue = '1.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000' AND outer shape = '0.055137 0.013605 0.033333 0.064517 0.041006 0.055137 0.055137 0.013605 0.033333 0.064517 0.041006 0.055137 0.055137 0.041033 0.05 0.055137 0.041006 0.064517 0.033333 0.013605 0.055137 '

(7) Inquiry to each DB: DB1: //hostname/DB1/picture/pict01.gif, “red ellipse” DB2: (binary data), “dart”

(8) Convert search result: DB2: The value of the image is converted because the user domain and the local domain are different. PictureData2Pictur
The result of eFileName ((binary data)) is '//hostname/DB2/picture/pict01.gif'

(9) Sum the search results. The final search result is //hostname/DB1/picture/pict01.gif, "red ellipse" //hostname/DB2/picture/pict01.gif, "dart"

[0052]

As described above, according to the present invention,
An image database in which search data of an inquiry sentence input by a user is recorded and a local domain corresponding to the search data are searched from a local domain management table, and a user domain is determined from a user name and a domain group of the search data designated by the user. Is searched from the user domain management table, and when the searched local domain is different from the user domain, a conversion function for converting the user domain to the local domain and a conversion function for converting the local domain to the user domain are stored in the conversion function management table. From the query database, using the former conversion function, convert the query sentence from the user into an expression format suitable for each image database, translate it into a query sentence for that image database, and use this translated query sentence to translate each image database. The search function is used to convert the expression form and delivery method of the search result into the user domain using the latter conversion function. Therefore, the image feature expression form specified in the search condition of each image database and the search result for each image database Even if the delivery method of the image data to the application program differs from the expression format or delivery method specified by the user, the image data can be easily obtained by searching freely, and the data in the image database can be used more freely. Can be. In addition, it is not necessary to define a global schema, and it is possible to eliminate differences in expression format and image data delivery method.

Further, in the existing multi-database, if the expression form of the data items in all the databases, the expression form of the image feature specified in the search condition, and the method of transferring the image data to and from the application program are not known, the difference is determined. In contrast to the above, according to the present invention, the administrator of the local database has a representation of data items in his / her own image database, a representation of an image feature specified in a search condition, and transfer of image data to and from an application program. It is only necessary to be aware of the local domain, which is the method, and it is not necessary to be aware of the expression format of other image databases or the delivery method of image data.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of an image database heterogeneity resolving search device according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram showing the concept of domain management.

FIG. 3 is a diagram showing a configuration of a column table, a domain table, a domain group table, and a domain group user table stored in a metadata storage unit used in the image database dissimilarity search device of FIG.

FIG. 4 is a diagram showing a configuration of a conversion function table stored in a metadata storage unit used in the image database dissimilarity search apparatus of FIG. 1;

FIG. 5 is a diagram showing a sample database.

FIG. 6 is a diagram showing an example of an image correspondence table used by a conversion function.

FIG. 7 is a diagram illustrating an example of a hue correspondence table used by a conversion function.

FIG. 8 is a diagram illustrating an example of a correspondence table of external shapes used by a conversion function;

FIG. 9 is a flowchart showing a schematic operation of the image database heterogeneity resolving and searching device of the present embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Syntax analysis part 2 Query processing control part 3 Query conversion part 4 Conversion function library 5 Database access part 6 Data processing part 7 Metadata storage part 8 Metadata management part 9 Image database 10 Database management system 11 Query processing part 12 Application program 13 Inquiry Statement 15 Database System

 ──────────────────────────────────────────────────の Continuing on the front page (72) Inventor Takashi Hoshino Nippon Telegraph and Telephone Corporation, 3-19-2 Nishishinjuku, Shinjuku-ku, Tokyo

Claims (3)

[Claims] In an environment in which a plurality of image databases exist, each image database can specify the same image feature as a search condition, but the expression form of the image feature that can be specified as a search condition differs for each image database, An image database heterogeneity elimination search method that realizes a search that eliminates a difference in expression form and a difference in an image data transfer method when a method of delivering image data serving as a search result to an application program differs for each database. , Manages a virtual column in which an existing column of each image database and an item that can be specified as a search condition but the corresponding column does not exist in the image database, and data that can be specified in a search request of each image database. Item expression format and images that can be specified as search conditions for each image database The domain that is the expression format of the features and the method of delivering image data that is the search result of each image database is managed as a local domain in the local domain management table for each image database, and the expression format and the delivery method are the same. Domains are managed as domain groups in the domain group management table, conversion functions for converting between domains are managed in the conversion function management table, and data representation formats used in search conditions specified by the user.
An expression form of the data of the search result requested by the user and a domain which is a method of transferring the data of the search result requested by the user are managed as user domains in the user domain management table for each user, and an inquiry sentence input by the user Searches the image database in which the search data is stored and the local domain corresponding to the search data from the local domain management table, and finds the user domain from the user name and the domain group of the search data specified by the user in the user domain management table. When the searched local domain and the user domain are different, a conversion function for converting the user domain to the local domain and a conversion function for converting the local domain to the user domain are searched from the conversion function management table. Using the conversion function searched for, the query sentence is converted into an expression format suitable for each image database, translated into a query sentence for the image database, and each image database is searched using the translated query sentence. A method for resolving dissimilarity in an image database characterized by the following. 2. In an environment in which a plurality of image databases exist, each image database can specify the same image feature as a search condition, but the expression format of the image feature that can be specified as a search condition differs for each image database. An image database dissimilarity search device that realizes a search that eliminates a difference in expression form and a difference in a method of delivering image data when a method of delivering image data serving as a search result to an application program differs for each database. , Manages a virtual column in which an existing column of each image database and an item that can be specified as a search condition but the corresponding column does not exist in the image database, and data that can be specified in a search request of each image database. Item expression format and images that can be specified as search conditions for each image database A local domain management table that manages, for each image database, a domain, which is a method of delivering image data as a search result of each image database as a feature expression form, and a target of the same expression form and delivery method for each image database. A domain group management table that manages domains as domain groups, a conversion function management table that manages conversion functions that convert between domains, a data representation format used in search conditions specified by a user,
A user domain management table that manages, for each user, a domain as a user domain, which is a representation format of search result data requested by the user and a method of delivering search result data requested by the user, and a query sent from the user. Searches the image database in which the search data is stored and the local domain corresponding to the search data from the local domain management table, and finds the user domain from the user name and the domain group of the search data specified by the user in the user domain management table. A search function for converting the user domain into a local domain and a conversion function for converting the local domain into a user domain when the local domain and the user domain searched by the domain search means are different. A conversion function search unit for searching from the conversion function management table; and a conversion function searched by the conversion function search unit, converting the query sentence into an expression format suitable for each image database, and inquiring the image database. A search means for translating into a sentence and searching each of the image databases with the translated query sentence. 3. In an environment in which a plurality of image databases exist, each image database can specify the same image feature as a search condition, but the expression format of the image feature that can be specified as a search condition differs for each image database. When the delivery method of image data as search results to application programs differs for each database, an image database heterogeneity resolution search program that realizes a search that eliminates differences in expression form and differences in image data delivery methods was recorded. A storage medium that manages a virtual column in which an existing column of each image database and a column that can be specified as a search condition but whose corresponding column does not exist in the image database, The expression form of data items that can be specified in The expression format of image features that can be specified as a condition and the domain that is the method of delivering image data that is the search result of each image database are managed as local domains in the local domain management table for each image database, and the expression format and delivery method Each domain having the same target is managed as a domain group in a domain group management table, a conversion function for converting between domains is managed in a conversion function management table, and a data expression format used in a search condition specified by a user,
An expression form of the data of the search result requested by the user and a domain which is a method of transferring the data of the search result requested by the user are managed as user domains in the user domain management table for each user, and an inquiry sentence input by the user Searches the image database in which the search data is stored and the local domain corresponding to the search data from the local domain management table, and finds the user domain from the user name and the domain group of the search data specified by the user in the user domain management table. When the searched local domain and the user domain are different, a conversion function for converting the user domain to the local domain and a conversion function for converting the local domain to the user domain are searched from the conversion function management table. Using the conversion function searched for, the query sentence is converted into an expression format suitable for each image database, translated into a query sentence for the image database, and each image database is searched using the translated query sentence. A recording medium on which an image database dissimilarity search program is recorded.