JPH11312171A - Image retrieving device, its method and computer-readable memory - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain an image retrieving device capable of easily specifying the degree of notice to a retrieving condition for retrieving image data, its retrieving method and a computer-readable memory. SOLUTION: Image data, its attribute information and an image featured value are correspondingly managed by using an image storing part 4 and an information managing data base DB7. The attribute information, the image featured value and the degree of notice to the attribute information and the image featured value are specified from an input part 1. The attribute information and the image featured value constituted of the specified retrieving condition are matched with the attribute information and the image featured value corresponding to the image data managed by the managing means. Image data which are to become the retrieved result based on the matching result are outputted to a display part 9.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention manages image data,
The present invention relates to an image search apparatus and method for searching for desired image data from the managed image data, and a computer-readable memory.

[0002]

2. Description of the Related Art Conventionally, for a plurality of attribute information and image feature amounts in image data, numerical values corresponding to respective weights are designated using a user interface.
There is an image search for searching for desired image data by matching the attribute information and the image feature amount with the attribute information and the image feature amount of the image data to be searched. Most of this type of image retrieval uses a multivariate analysis mainly for statistical processing as a matching method to perform matching in consideration of a distance in a certain metric space, and at that time, coordinates relating to attribute information and image feature amount This was realized by component weighting.

As described above, in an image search method in which a plurality of pieces of attribute information and image feature values are designated with numerical values corresponding to their respective weights using a user interface, a matching method is performed in a certain image feature space. Almost the same approach was taken in that the geometric distance of the weighting was used.

Further, when matching attribute information and image feature amounts of image data to be searched using each attribute information and image feature amount, ambiguity of matching is determined for each attribute information and image feature amount. There are things that can be introduced.

[0005]

However, in the above-described conventional image retrieval, specifying the ambiguity in the matching of each attribute information and image feature amount separately from the numerical value corresponding to the weighting has a correct meaning for the user. It was hard to understand and inconvenient.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and has an image search apparatus and method capable of easily designating a degree of attention to search conditions for searching image data, and a computer-readable memory. The purpose is to provide.

[0007]

The image retrieval apparatus according to the present invention for achieving the above object has the following arrangement.
That is, an image search device that manages image data and searches for desired image data from the managed image data, and a management unit that manages the image data in association with the attribute information and the image feature amount, Specifying means for specifying the attribute information and the image feature amount as search conditions for searching for image data, and a degree of attention to the attribute information and the image feature amount; and attribute information configured by the search condition specified by the specifying means; A matching unit that matches the image feature amount and the corresponding attribute information and image feature amount of the image data managed by the management unit based on the degree of interest; and a search result based on the matching result by the matching unit. And output means for outputting image data.

Preferably, the designating means has a user interface for designating a numerical value corresponding to the degree of interest, and the matching means is used when matching the degree of interest specified by the user interface. The attribute information is converted into an ambiguity, and the attribute information and the image feature amount included in the search condition are matched with the corresponding attribute information and the image feature amount of the image data by using the evaluation expression based on the ambiguity.

Preferably, the degree of attention is a weight corresponding to the importance for the attribute information and the image feature amount, and the lower the value of the degree of attention, the greater the ambiguity of matching by the matching means. It is to let.

Preferably, the output means sorts and outputs image data as a search result based on the degree of coincidence of image feature amounts obtained as a result of matching by the matching means.

Preferably, the degree of coincidence is a normalized degree of coincidence divided by a total sum of degrees of interest of the attribute information and image feature amounts.

[0012] Preferably, the matching means includes an image feature amount of image data having a degree of coincidence equal to or more than a predetermined value for each image feature amount obtained by digitizing the attribute information and the image feature amount included in the search condition. An extraction unit for extracting a feature amount in advance is provided, and the image feature amount of the image data extracted by the extraction unit is matched with the image feature amount included in the search condition.

Preferably, for the unextracted image data not extracted by the extracting means, the lowest coincidence of the image data extracted by the extracting means is determined to be the coincidence of the unextracted image data. Degree.

An image search method according to the present invention for achieving the above object has the following arrangement. That is, an image search method for managing image data and searching for desired image data from the managed image data, wherein the image data is associated with its attribute information and image feature amount and managed in a storage medium A process for specifying the attribute information and the image feature amount as search conditions for searching for image data, and a degree of attention to the attribute information and the image feature amount, and a search condition specified in the specifying step Attribute information and image feature amount, the corresponding attribute information and image feature amount of the image data managed in the storage medium in the management step,
A matching step of matching based on the degree of interest; and an output step of outputting image data as a search result based on a matching result of the matching step.

A computer readable memory according to the present invention for achieving the above object has the following configuration. That is,
A computer-readable memory storing image search program codes for managing image data and searching for the desired image data from the managed image data, wherein the image data is associated with attribute information and image feature amounts. A program code of a management step of managing the storage medium and the program code of a designation step of designating the attribute information and the image feature amount as search conditions for searching for image data, and a degree of attention to the attribute information and the image feature amount. The attribute information and the image feature amount configured by the search condition specified in the specifying step, and the corresponding attribute information and image feature amount of the image data managed in the storage medium in the management step, Based on the program code of the matching step for performing matching based on the And a program code for an output step of outputting the image data to be the result.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a block diagram showing a configuration of an image retrieval apparatus according to an embodiment of the present invention.

In FIG. 1, reference numeral 1 denotes an input unit, which comprises a keyboard and a mouse. 4 is an image storage unit,
The image data temporarily stored in the image memory 5 is stored.
An image input unit 2 inputs image data to be registered in the image storage unit 4. An image memory 5 temporarily stores image data input from the image input unit 2.
Reference numeral 6 denotes an image feature amount extraction unit which acquires attribute information indicating attributes (image, text, etc.) of the input image data and digitizes the attribute information. Further, the image feature amount of the image data is extracted and normalized. In addition, this normalization is generally performed on the calculated image feature amounts under the same condition when performing a matching process and discussing a distance obtained by integrating data having completely different characteristics. To do that. The data (average,
(Distribution) is stored in the storage area of the system at the time of introduction, and has a configuration as shown in FIG. 10, for example.

Reference numeral 7 denotes an information management database (DB) for registering attribute information and image feature amounts of the image data extracted by the image feature amount extraction unit 6 in association with the image data. The attribute information may be stored separately in the management database (DB) 8. Reference numeral 3 denotes a matching unit that matches attribute information and image feature amounts included in the search condition with corresponding attribute information and image feature amounts of the search target image data, and obtains a matching degree. A display unit 9 displays image data obtained as a search result from the matching unit 3. In addition, a user interface for designating the degree of interest of the search condition to the user is displayed.

First, a registration process for registering image data will be described with reference to FIG.

FIG. 2 is a flowchart showing a registration process of image data executed in the embodiment of the present invention.

First, in step S101, attribute information of input image data is obtained. In step S102,
Digitize the acquired attribute information. This digitization is performed using a predetermined calculation formula or a predetermined conversion table. In step S103, the image feature amount of the input image data is calculated. In step S104, data (average, variance) for normalizing image data from the storage area of the system
Is acquired, and the image feature amount is normalized based on the acquired data. In step S105, for example, in the configuration shown in FIG. 3, the information ID indicating the input image data
, The normalized image feature amount, the acquired attribute information,
The image data is stored in the information management DB 7 in association with an information storage destination address indicating a place where the image data is actually managed.

Next, a search process for searching for image data will be described with reference to FIG.

FIG. 4 is a flowchart showing image data search processing executed in the embodiment of the present invention.

First, in step S201, attribute information and an image feature amount are specified as search conditions, and their attention degrees are specified.

The designation here is "it is reasonable to perform a matching process by increasing the degree of ambiguity for those having a low numerical value (degree of attention) corresponding to the weighting of the attribute information or image feature amount. Yes and not against the user's will "
This is based on the idea. Thus, the degree of attention can specify the weighting of the attribute information and the image feature amount, and can specify the ambiguity of the matching process of each attribute information and the image feature amount. In other words, the attribute information and the image feature amount with a larger weighting reduce the ambiguity at the time of matching, and conversely, the attribute information and the image feature amount with a smaller weighting increase the ambiguity at the time of matching.

In order to realize the above relationship, if the degree of ambiguity is discrete, the "correspondence function between the degree of interest and the degree of ambiguity" shown in FIG. 5 is used. If the degree of ambiguity is continuous, a “monotone decreasing function representing the correspondence between the degree of attention and the degree of ambiguity” shown in FIG. 6 is used. However, the contents of the functions in FIGS. 5 and 6 are not particularly limited as long as they are monotonically decreasing functions.

As the search condition, coordinates on the normalized multi-dimensional image feature space at the time of registration and at the time of registration are given. For example, when searching for image data similar to a certain image, attribute information of the image data and a normalized image feature amount are given.

In step S202, based on the specified degree of attention, the attribute information and the normalized image feature amount included in the specified search condition (hereinafter, these may be collectively referred to as normalization parameters). And the normalization parameter of the first image data stored in the image storage unit 4 by the matching unit 3. Specifically, the matching between the normalized parameters is performed in accordance with the degree of ambiguity obtained based on the degree of attention, and the matching result is set as the similarity of the normalized parameter of interest. In addition, the calculation formula of the similarity can be changed according to the degree of ambiguity, or the degree of restriction can be adjusted. Then, using the obtained similarity for each normalization parameter, the degree of coincidence L between the normalization parameter of the image data to be processed and the normalization parameter included in the search condition is calculated. The calculation formula for calculating the coincidence L can be expressed as follows.

[0030]

(Equation 1)

Where i: identifier of image feature amount k: number of types of normalization parameters P _i : normalization parameter R _i : comparison reference parameter (normalization parameter of image data to be matched) W _i : normalization parameter i (F _i (W _i )): A function for converting the degree of attention to the image feature i into ambiguity S _i (F _i , R _i , P _i ): Ambiguity F to the image feature i
Function for Obtaining Similarity between Comparison Reference Parameter R _i and Normalized Parameter P _i Considering _i Here, the details of the process of obtaining the degree of coincidence L in step S202 will be described with reference to FIG.

FIG. 7 is a flowchart showing details of the processing executed in step S202 of FIG. 4 according to the embodiment of the present invention.

First, in step S301, a normalization parameter P _i , a degree of interest W _i , and the number k of types of the normalization parameter, which constitute a search condition, are obtained. In step S302,
The value of the counter i for counting the normalized parameters for which matching has been completed is set to 1, and the value of the register L (degree of coincidence) is set to 0. In step S303, the counter i
Is determined to be less than or equal to the number k of types of normalization parameters. If the value of the counter i is equal to or smaller than the number k of types of the normalization parameters (YES in step S303), the process proceeds to step S304, and focuses on the normalization parameters P _{i i} using F _i = F _i (W _i ). to convert degrees Wi to the ambiguity of F _i. In step S305, the normalization parameter P _i and the comparison reference parameter R _i are matched according to the ambiguity F _i to obtain a similarity S = S _i (F _i , R _i , P _i ). And
The obtained similarity S is added to the current value of the register L.

On the other hand, when the value of the counter i is larger than the number k of types of normalization parameters (N in step S303)
O), the process proceeds to step S307, and the sum of the similarities S _i is divided by the value of the register L to obtain a normalized coincidence L.
In step S308, the obtained matching degree L and the information ID of the image data to be matched are paired and output as a search result.

Returning to the description of FIG.

In step S203, it is determined whether there is no matching image data. If there is no matching-target image data (YES in step S203), the process proceeds to step S207, and the information ID of the image data and the matching degree L between the search conditions are output as a search result to the display unit 9 as a pair. On the other hand, when image data to be matched exists (NO in step S203), the process proceeds to step S204, and the next image data to be matched is selected.

In step S205, the matching unit 3 matches the normalization parameter included in the search condition with the normalization parameter of the image data. The details of this processing are the same as in step S202, and thus are omitted here. In step S206, the image data for which matching by the matching unit 3 has been completed is sorted based on the obtained matching degree L, and the process returns to step S203. Then, in step S203, when the image data to be matched no longer exists, the process proceeds to step S207, and the information ID of the image data and the degree of matching L with the search condition are displayed as a search result in descending order of the degree of matching L. Output to the unit 9.

In the search processing described above,
In order to shorten the time, pre-search may be performed
It is possible. That is, for each image feature amount,
Extract image data with criticality in advance and
Search based on the search conditions actually specified
Perform search processing. However, in this case, all the normalized parameters
Degree S_i(W_i) Is not obtained. There
In such a case, the regular
Similarity S corresponding to the optimization parameter i_i(F_i, R _i, P_i)
For, the most similarity of the search results remaining in the pre-search
Min (S_i) To S_i(F_i, R_i, P_i)
To use. FIG. 8 shows a search process including this pre-search.
This will be described with reference to FIG.

FIG. 8 is a flowchart showing a search process including a pre-search executed in the embodiment of the present invention.

First, in step S401, attribute information and an image feature amount are specified as search conditions, and their attention degrees are specified. In step S402, as a pre-search, the matching unit 3 performs matching with the normalization parameter of the image data stored in the image storage unit 4 for each normalization parameter included in the selected search condition. And
For each image data, the presence / absence of the pre-search result, the information ID, and the similarity are stored, and the minimum value of the similarity for each normalization parameter is stored.

Next, in step S403, the image storage unit 4
Is calculated by the above (Equation 1.0) for each normalization parameter of the first image data stored in the first image data, and the degree of coincidence L with the search condition is obtained. However, if the pre-search of the normalization parameter j is unsuccessful, the minimum value of the similarity of the normalization parameter j stored in step S402 for the term of i = j in (Equation 1.0) Substitute This makes it possible to compensate for missing calculation results of the pre-search for each normalization parameter and perform matching.

In step S404, it is determined whether there is no image data to be processed for which the calculation of (Expression 1.0) is performed. If there is no image data to be processed (YES in step S404), the process advances to step S408 to output the information ID of the image data and the degree of coincidence L between the search conditions to the display unit 9 as a search result. On the other hand, if there is image data to be processed (NO in step S404),
Proceeding to step S405, the next image data to be processed is selected.

In step S406, the above-described (Equation 1.0) is calculated for each normalization parameter of the image data to obtain a degree of coincidence L with the search condition. Step S4
In step 07, the processed image data is sorted based on the obtained matching degree L, and the process returns to step S404. Then, in step S404, if there is no more image data to be processed, the process proceeds to step S408, and the information ID of the image data and the degree of matching L with the search condition are displayed as a search result in descending order of the degree of matching L. Output to the unit 9.

Next, specific examples of the above-described registration processing and search processing will be described.

Here, as the image data to be registered in the registration processing, an image data represented by a 32-bit RGB value, particularly, a figure having a figure painted in one color on a white background will be described as an example.

In the registration process, as described above, first, the attribute information of the image data to be registered is obtained and the image feature amount is calculated. In this case, in order to convert the attribute information into a numerical value and obtain an image feature amount, the color indicated by the RGB value of the image data is converted into a numerical value using the conversion table shown in FIG. Also,
The area of the composition indicated by the filled figure included in the image data is obtained, and the obtained area is regarded as the image feature.

Next, data for normalizing these image feature values is obtained from the table shown in FIG. 10, and a normalized image feature value X is obtained according to the following equation.

Normalized image feature amount X = (image feature amount−average) / variance (Equation 2.0) Then, the image feature amount obtained by digitizing the color represented by the actual RGB value is represented by a composition. When the image features are normalized, they are stored in the information management DB 7.

Subsequently, as the search conditions, the degree of attention of the image feature amount represented by the color is 70%, the degree of attention of the image feature amount represented by the area of the composition is 30%, and any image data from the information management DB 7 is retrieved. The search process in the case where it is specified will be described. It is assumed that, of the image feature amounts of the designated image data, the image feature amount obtained by color is 65, and the image feature amount indicated by the composition area is 20. In addition, the degree of attention of the image feature amount represented by a color is converted into an ambiguity using a discrete function shown in FIG. Further, the degree of attention of the image feature amount represented by the area of the composition is converted into the degree of ambiguity using a discrete function shown in FIG. In this case, the ambiguity of the image feature indicated by the color is 10, and the ambiguity of the image feature related to the area of the composition is 80.

When the degree of coincidence with the search target image is calculated according to the above (Equation 1.0) using the above results, the number k of types of image feature amounts (normalization parameters) is k = 2. Equation 1.0) can be written as:

L = 70 × S ₁ (10,65, P ₁ ) + 30 × S ₂ (80,20, P ₂ )｝ (Expression 3.0) Here, the first term of (Expression 3.0) ( S ₁ (10,65,
P ₁ )) is used in pattern matching for color.
This means that the calculation of the similarity between the image feature quantity 65 and P1 relating to the ambiguity 10 is calculated. Similarly, (Equation 3.0)
The second term (S ₂ (80, 20, P ₂ )) means that, in pattern matching relating to the area of the composition, calculation of the similarity between the image feature quantity 20 relating to the ambiguity 80 and P 1 is calculated.

In the calculation of the degree of similarity, an arithmetic expression / operation algorithm suitable for the degree of ambiguity is used for each image feature amount.

The above processing is performed on all image data to be searched, and the image data is sorted in descending order of the obtained coincidence L, and is displayed on the display unit 9 as a search result. still,
When the search result is returned to a higher-level application, or when the distance is indicated by a distance instead of the degree of coincidence L, the conversion may be performed through a function that associates the distance with the degree of coincidence. It becomes a decreasing function.

When the search processing including the pre-search is performed, if the difference between the color of the image feature represented by the color and the color of the image feature corresponding to the image data to be processed is within a predetermined range, Alternatively, if the ratio of the composition area of the image feature amount represented by the composition area to the corresponding image feature amount of the image data to be processed is within a predetermined range, the pre-search is passed for each image feature amount. And Then, for each of the image data to be processed, the presence / absence of the pre-search result, the information ID, and the similarity are stored, and the minimum value of the similarity for each image feature amount (normalization parameter) is stored.
That is, the minimum value of the first and second terms of (Equation 3.0) is stored.

Then, the calculation based on (Equation 3.0) is performed for all the image data, and the degree of coincidence L with the search condition is obtained. In this case, the image data is sorted in descending order of the matching degree L. In the image data in which the pre-search of the image feature amount represented by the color has failed, the first term in (Equation 3.0) is compared with the first term. Is substituted for the minimum value of the first term. Further, in the image data in which the pre-search of the image feature amount indicated by the composition is rejected, (Expression 3.0)
Substitute the minimum value of the second term previously stored for the second term in.

As described above, according to the present embodiment, it is possible to specify the degree of attention to the image feature amount suitable for the search purpose, and to convert the specified degree of attention to the degree of ambiguity. It is possible to set the ambiguity of the image feature amount suitable for the purpose.

In the above embodiment, the normalization parameter of the search condition and the normalization parameter of the image data to be searched are matched with each other to obtain the degree of coincidence L. May be subjected to multivariate analysis to determine the principal component axis, reduce the dimension to the number of effective principal components, and perform matching of the reduced dimension on the principal component space. In that case, the ambiguity of the image feature amount (normalization parameter) must be determined as depending on the principal component axis. In many cases, the principal component axis is often an axis having a certain meaning physically and conceptually, and such weighting of the principal component axis becomes more effective than in the present embodiment.

In the above embodiment, the degree of coincidence L between the normalization parameter included in the search condition and the normalization parameter of the image data to be searched is determined by using the similarity (Equation 1.0). Of course, an algorithm that focuses on the distance is also conceivable. In that case, the formula for calculating the total distance Diff corresponding to (Formula 1.0) is as follows.

[0059]

(Equation 2)

Where i: identifier of image feature amount k: number of types of normalization parameters P _i : normalization parameter R _i : comparison reference parameter (normalization parameter of image data to be matched) W _i : normalization parameter i degree of attention F _i (W _i) with respect to: an image feature amount function for converting the degree of attention to the ambiguity for _{i D i (F i, R} i, P i): ambiguity F for the image feature amount i
function to obtain the similarity between the comparison reference parameters R _i and normalized parameters P _i in consideration of _i However, since the distance Diff is closely related to the inverse proportion to the degree of coincidence L, when performing the processing described above with reference to this Need to take this into account.

Even if the present invention is applied to a system including a plurality of devices (for example, a host computer, an interface device, a reader, a printer, etc.), a device including one device (for example, a copying machine, a facsimile machine) Etc.).

Another object of the present invention is to provide a storage medium storing a program code of software for realizing the functions of the above-described embodiments to a system or apparatus, and to provide a computer (or CPU) of the system or apparatus.
And MPU) read and execute the program code stored in the storage medium.

In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiment, and the storage medium storing the program code constitutes the present invention.

As a storage medium for supplying the program code, for example, a floppy disk, hard disk, optical disk, magneto-optical disk, CD-ROM, CD
-R, a magnetic tape, a nonvolatile memory card, a ROM, or the like can be used.

When the computer executes the readout program code, not only the functions of the above-described embodiment are realized, but also the OS (Operating System) running on the computer based on the instruction of the program code. ) May perform some or all of the actual processing, and the processing may realize the functions of the above-described embodiments.

Further, after the program code read from the storage medium is written into a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer, based on the instruction of the program code, It goes without saying that the CPU included in the function expansion board or the function expansion unit performs part or all of the actual processing, and the processing realizes the functions of the above-described embodiments.

[0067]

As described above, according to the present invention,
It is possible to provide an image search device and a method thereof, and a computer-readable memory, which can easily designate a degree of attention to search conditions for searching image data.

[0068]

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an image search device according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a registration process of image data executed in the embodiment of the present invention.

FIG. 3 is a diagram showing a data configuration recorded in an information management DB according to the embodiment of the present invention.

FIG. 4 is a flowchart illustrating image data search processing executed in the embodiment of the present invention.

FIG. 5 is a diagram illustrating an example of a discrete function expressing a correspondence between a degree of interest and an ambiguity according to the embodiment of this invention;

FIG. 6 is a diagram illustrating an example of a continuous function expressing a correspondence between a degree of interest and an ambiguity according to the embodiment of this invention;

FIG. 7 is a flowchart showing details of processing executed in step S202 of FIG. 4 according to the embodiment of the present invention.

FIG. 8 is a flowchart illustrating a search process including a pre-search executed in the embodiment of the present invention.

FIG. 9 is a diagram illustrating an example of a table for digitizing attribute information according to the embodiment of this invention.

FIG. 10 is a diagram showing an example of data for normalizing an image feature amount according to the embodiment of the present invention.

FIG. 11 is a diagram illustrating an example of a discrete function expressing a correspondence between a degree of attention and an ambiguity of an image feature amount indicated by a color according to the embodiment of this invention.

FIG. 12 is a diagram illustrating an example of a discrete function expressing a correspondence between a degree of interest and an ambiguity of an image feature amount indicated by a composition area according to the embodiment of this invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Input part 2 Image input part 3 Matching part 4 Image storage part 5 Image memory 6 Image feature quantity extraction part 7 Information management database 8 Management database 9 Display part

Claims (15)

[Claims] 1. An image search apparatus for managing image data and searching for desired image data from the managed image data, wherein the image data is managed in association with attribute information and image feature amounts. Means, designating means for designating the attribute information and the image feature amount, and a degree of attention to the attribute information and the image feature amount as search conditions for searching for image data; and a search condition designated by the designating means. A matching unit that matches the attribute information and the image feature amount and the corresponding attribute information and image feature amount of the image data managed by the management unit based on the degree of interest; and based on a matching result by the matching unit. ,
Output means for outputting image data as a search result. 2. The method according to claim 1, wherein the specifying unit has a user interface that specifies a numerical value corresponding to the degree of interest, and the matching unit determines an ambiguity when the degree of interest specified by the user interface is matched. And converting the attribute information and the image feature amount included in the search condition with the corresponding attribute information and image feature amount of the image data using the evaluation formula based on the degree of ambiguity. 2. The image search device according to 1. 3. The degree of interest is a weight corresponding to the degree of importance for the attribute information and the image feature amount, and the ambiguity of matching by the matching means increases as the value of the degree of interest decreases. 2. The image search device according to claim 1, wherein: 4. The image processing apparatus according to claim 1, wherein the output unit sorts and outputs image data as a search result based on a degree of coincidence of an image feature amount obtained as a matching result by the matching unit. The image search device according to the above. 5. The image search apparatus according to claim 4, wherein the degree of coincidence is a normalized degree of coincidence divided by a total sum of degrees of interest of the attribute information and the image feature amount. 6. The image processing apparatus according to claim 1, wherein the matching unit determines an image feature amount included in the search condition and an image feature amount of image data having a matching degree equal to or more than a predetermined value for each image feature amount obtained by digitizing the attribute information. An extraction unit that extracts in advance, an image feature amount of the image data extracted by the extraction unit,
The image search device according to claim 1, wherein the image feature amount configured by the search condition is matched. 7. Regarding the unextracted image data not extracted by the extracting unit, the lowest coincidence of the image data extracted by the extracting unit is defined as the coincidence of the unextracted image data. 7. The image search device according to claim 6, wherein: 8. An image retrieval method for managing image data and retrieving desired image data from the managed image data, wherein the image data is associated with attribute information and image feature amounts in a storage medium. A management step of managing; a designation step of designating the attribute information and the image feature amount as search conditions for retrieving image data; and a degree of interest in the attribute information and the image feature amount; and a search condition designated in the designation step. A matching step of matching, based on the degree of attention, the attribute information and the image feature amount configured by the attribute information and the corresponding attribute information and image feature amount of the image data managed in the storage medium in the management step; Based on the matching result of the process,
Outputting an image data as a search result. 9. The designating step includes a user interface for designating a numerical value corresponding to the degree of interest, and the matching step includes determining a degree of ambiguity when matching the degree of interest designated by the user interface. And converting the attribute information and the image feature amount included in the search condition with the corresponding attribute information and image feature amount of the image data using the evaluation formula based on the degree of ambiguity. 8. The image search method according to 8. 10. The degree of interest is a weight corresponding to the degree of importance for the attribute information and the image feature amount, and at the same time, the lower the value of the degree of interest, the greater the ambiguity of matching in the matching step. 9. The image search method according to claim 8, wherein: 11. The image processing apparatus according to claim 8, wherein the output step sorts and outputs image data as a search result based on the degree of coincidence of image feature amounts obtained as a result of the matching in the matching step. Image search method described. 12. The image search method according to claim 11, wherein the degree of coincidence is a normalized degree of coincidence divided by a total sum of degrees of interest of the attribute information and the image feature amount. 13. The image processing apparatus according to claim 1, wherein the matching step includes: determining an image feature amount of the image data having a degree of coincidence equal to or more than a predetermined value for each image feature amount obtained by digitizing the attribute information. An extraction step for extracting in advance, an image feature amount of the image data extracted in the extraction step,
9. The image search method according to claim 8, wherein an image feature amount included in the search condition is matched. 14. Regarding unextracted image data not extracted in the extraction step, the lowest coincidence among the coincidences of the image data extracted in the extraction step is defined as the coincidence of the unextracted image data. 14. The image search method according to claim 13, wherein: 15. A computer-readable memory storing image search program codes for managing image data and retrieving desired image data from the managed image data, comprising: image data, its attribute information, and image characteristics. A program code of a management step of managing the amounts in the storage medium in association with the amount, and designation of designating the attribute information and the image feature amount as search conditions for searching for image data, and a degree of attention to the attribute information and the image feature amount A program code of a step, attribute information and an image feature amount included in the search condition specified in the specifying step, and corresponding attribute information and image feature amount of image data managed in the storage medium in the management step. A program code of a matching step for matching based on the degree of interest; and matching by the matching step. Based on the results,
And a program code of an output step of outputting image data as a search result.