JP2012003357A - Feature information creation device, method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent deterioration of retrieval accuracy when similar image retrieval is performed with an image as a key.SOLUTION: An existing web retrieval engine E is used to obtain an image from a web server GS on the basis of a retrieval result obtained by providing a keyword for the web retrieval engine. An image feature amount extracted from the image is stored and accumulated for use in similar image retrieval, so that a similar image retrieval engine is realized without a large burden associated with specific use of a crawler. Deterioration of retrieval accuracy that occurs when an irrelevant image similar only in appearance hits as a similar image candidate during a retrieval using only image feature amount is avoided by storing text feature amount in addition to image feature amount for consideration into a similar image retrieval.

Description

  The present invention relates to a similar image search using an image as a key.

  2. Description of the Related Art Conventionally, with the enrichment and diversification of content on the web, methods for searching various information published on the Internet are increasing. As a basic method, keyword search using a specified keyword as a search key (query) is known, and in addition, an image specified by a search (called “query image” or the like) is used as a search key and similar to it. A so-called similar image search in which an image is searched and displayed is known (for example, see Non-Patent Document 1).

  In the similar image search, images similar to the query image are searched and displayed by comparing information called feature amounts obtained by converting visual features such as color and shape of each image into data. Here, as a search result candidate, a large number of images to be compared with the query image are hereinafter referred to as “candidate images”. In order to implement the similar image search as described above in the system or the like, It is necessary to prepare in advance as a database or the like by associating a large number of candidate images with feature amounts representing colors and shapes extracted from the image data.

Google Inc., “Google Similar Images”, [online], [Search March 17, 2010], Internet <URL: http://similar-images.googlelabs.com/>

JP 2008-160556 A

  Here, as an example of a conventional technique for registering an image in relation to a search, there is an example in which an image related to a keyword input by a user is searched by an Internet search engine and a hit image is registered (see Patent Document 1). Images that can be acquired by this type of search are thumbnail images and cache images, and original image data suitable for feature amount extraction for similar image search cannot be acquired.

  In this regard, as a technique for comprehensively obtaining arbitrary data from the web, there is a crawling by a crawler. Crawling is a process in which a computer program called a crawler accesses and collects information by crawls on the web based on hyperlinks between web pages, URL hierarchical structures, and the like.

  Although it is theoretically possible to collect images by this crawling, crawl schedule management, indexing processing for generating search data (so-called index data) from collected information, ranking processing for determining priority among information, etc. The creation and operation of crawlers are costly. For this reason, conventionally, it has not been easy to implement a similar image retrieval system that enables retrieval of images on the web.

  In view of the above problems, an object of the present invention is to create a similar image search system by creating similar image search data without a large load of preparing a crawler independently.

  Based on the above object, one aspect (1) of the present invention is a feature information creation device that creates feature storage means for storing image feature quantities for each image in order to search for an image similar to a designated image. The search access means for transmitting the keyword to the web search engine and receiving the search result related to the transmitted keyword, and the image by accessing the web server based on the identification information of the information included in the received search result. Image acquisition means for acquiring data, and feature quantity extraction means for extracting an image feature quantity from the acquired image data and storing the image feature quantity in association with the identification information in a predetermined feature storage means It is characterized by that.

  According to another aspect (5) of the present invention, the above aspect is captured by a method category, and a feature storage means for storing an image feature amount for each image in order to search for an image similar to a designated image is provided. A method of creating characteristic information created by a computer, wherein the computer transmits a keyword to a web search engine and receives a search result relating to the transmitted keyword, and information included in the search result received by the computer An image acquisition step of acquiring image data by accessing a web server based on the identification information of the image, and the computer extracts an image feature amount from the acquired image data, and the image feature amount corresponds to the identification information And a feature amount extraction step for storing the feature amount in a predetermined feature storage means.

  According to another aspect (6) of the present invention, the above aspect is captured by a category of a computer program, and a feature memory that stores an image feature amount for each image in order to search for an image similar to a designated image. A feature information creation program for causing a computer to create means, which controls a computer to cause a web search engine to transmit a keyword, to receive a search result related to the transmitted keyword, and to be included in the received search result The image data is acquired by accessing the web server based on the identification information of the information to be extracted, the image feature amount is extracted from the acquired image data, and the image feature amount is associated with the identification information to obtain a predetermined feature It is stored in a storage means.

  In this way, utilizing an existing web search engine, acquiring an image from a web server based on a search result obtained by passing a keyword to the web search engine, storing and accumulating image features extracted from the image, and similar images By using it for the search, a similar image search engine can be realized without a large load using a crawler independently.

  According to another aspect (2) of the present invention, in any one of the above aspects, the web search engine is an image search engine that searches for an image based on a keyword.

  As described above, since an image is obtained as a search result by using an image search engine that searches for an image from a keyword as the web search engine, an original image is provided as compared with the case of obtaining a web page as a search result. The web server can be accessed quickly and processing can be made more efficient.

  According to another aspect (3) of the present invention, in any one of the above aspects, the feature amount extraction unit further includes a text feature amount based on a keyword transmitted to the web search engine by the search access unit, and the image feature. It is characterized by being stored in the feature storage means in association with the quantity or the identification information.

  In this way, by storing not only the image feature amount but also the text feature amount and adding to the similar image search, another similar image candidate similar only in appearance is hit in the case of only the image feature amount, and the search accuracy is deteriorated. The problem can be avoided.

  According to another aspect (4) of the present invention, in any one of the above aspects, for the image specified in the similar image search request, whether or not the image feature amount is stored in the feature storage unit is determined. For the image determined not to be stored, the image acquisition unit acquires image data, the image feature amount is extracted by the feature amount extraction unit, and the image feature amount is associated with the identification information. It is characterized by having a discrimination control means for storing in the feature storage means.

  As described above, when the image feature amount of the image specified in the similar image search is not stored, the image is acquired, and the image feature amount is extracted and stored. Sequential enhancements can improve the quality of similar image retrieval.

  It should be noted that a category (method for the apparatus, program for the method, etc.) different from each of the above-described modes and more specific modes described below are also included in the present invention. For different categories, “means” shall be appropriately read as “step”.

  According to the present invention, it is possible to prevent a decrease in search accuracy when a similar image search is performed using an image as a key.

The functional block diagram which shows the structure of embodiment of this invention. The figure which illustrates the information (data) used by embodiment of this invention. The flowchart which shows the process sequence of the feature information creation in embodiment of this invention. The flowchart which shows the process sequence of the similar image search in embodiment of this invention. The figure which shows the example of a screen for searching an image from a text in embodiment of this invention. The figure which shows the example of a screen which selects an image for a similar image search in embodiment of this invention. The conceptual diagram which shows an example of the weighting in embodiment of this invention. The conceptual diagram which shows an example of the weighting in embodiment of this invention. The figure which shows the example of a screen display of the similar image search result in embodiment of this invention. The figure which shows the example of a screen display of the similar image search result in embodiment of this invention. The figure which shows the example of a screen display of the similar image search result in embodiment of this invention.

  Next, modes for carrying out the present invention (referred to as “embodiments”) will be described with reference to the drawings. It should be noted that assumptions common to those already described in the background art and problems are omitted as appropriate.

[1. Constitution〕
The present embodiment is an example in which the feature information creation apparatus of the present invention is applied to the search server 1 in FIG. 1 (hereinafter referred to as “server” or “server 1”). A feature storage unit 21 that stores image feature quantities for each image and a feature storage unit 21 that is used to search for an image similar to the image specified via the communication network N. And a function of providing a similar image search service in response to a request from the terminal T.

  The server 1 has, as a general computer configuration, at least an arithmetic control unit 6 such as a CPU, a storage device 7 such as an external storage device (HDD or the like) or a main memory, and a communication network N (Internet, mobile phone network, LAN, etc.). Communication means 8 (such as a LAN adapter and a communication gateway). Although two communication networks N are shown in FIG. 1, they may be the same or different from each other. The terminal T is an information processing apparatus used as a client system by a user, such as a personal computer (PC), a smart phone, or a mobile phone terminal device. FIG. 1 is a schematic one, and there are actually many.

  In the server 1, elements (11, 12, and so on) shown in FIG. 1 are respectively controlled by a predetermined computer program (not shown) stored (installed) in advance in the storage device 7 by controlling the arithmetic control unit 6. 13 ..., 21, 22, 23 ... etc.). Among these elements, the information storage means can be realized in the storage device 7 in any format such as various files and databases (also referred to as “DB”), variables such as arrays, various stacks and registers, and system setting values.

  Among such storage means, the feature storage means 21 is a means for storing an image feature amount for each candidate image to be searched in order to search for an image similar to the designated query image (for example, FIG. 2. The configuration of the server 1 as a feature information creation device in the server 1 stores such image feature quantities in the feature storage unit 21. Note that the configuration of the feature information creation device and the configuration of the similar image search device that searches for similar images are not limited to being integrated, and may be different. Each means other than the storage means is a processing means for realizing and executing the following information processing functions and operations.

[2. (Feature information creation)
In the present embodiment configured as described above, a keyword is transmitted to a predetermined web search engine E determined in advance, and the feature storage unit 21 is created based on the obtained search result. As keywords used in this case, words frequently used in a search engine or words frequently used in a web page are prepared in advance as a keyword list D or the like, and the contents are sequentially processed. A flow chart of FIG. 3 shows a procedure of processing for creating the feature storage means 21 by storing feature quantities using such a keyword list D (referred to as “feature information creation”).

[2-1. (Acquiring images based on search results)
That is, if an unprocessed keyword remains in the keyword list D (step S10), the search access unit 12 extracts one unprocessed keyword from the keyword list D and transmits it to the web search engine E as a search request. The search result relating to the transmitted keyword is received (step S11). As the web search engine E, in the present embodiment, an existing image search engine that searches for an image based on a keyword is used.

  This web search engine E is a service realized on hardware such as a server, and crawls a web space such as the WWW (World Wide Web) to search related keywords for each keyword. Index data is stored in advance in the index storage means 11, and the search result of the image corresponding to the keyword specified in the search request from the access source is transmitted to the access source as a web page.

  The image acquisition unit 13 is a web server that provides an image based on identification information (for example, URL and file name for each image) of information included in the search result received by the search access unit 12. Image data is acquired by accessing the image server GS (step S12). For example, the image acquisition unit 13 sequentially acquires http files one by one to URLs corresponding to the images listed as thumbnails on the search result screen, thereby acquiring each image file as image data.

[2-2. Feature extraction)
Subsequently, the feature amount extraction unit 14 extracts the image feature amount from the image data acquired by the image acquisition unit 13, and stores the image feature amount in association with the identification information in the predetermined feature storage unit 21 (step S13). ). Here, the image feature amount is data representing visual features such as colors and shapes included in the image (may be contour lines, patterns, patterns, etc.), and the actual elements and expression formats are arbitrary, but for explanation. If a simplified and schematic example is shown, as illustrated in the column of “image feature amount” in FIG. 2, elements such as shapes and colors included in the image are converted into data in descending order or in order of conspicuousness. is there.

  In the present embodiment, the feature quantity extraction unit 14 uses the text feature quantity based on the keyword text transmitted to the web search engine E by the search access unit 12 as the text feature quantity based on the text (character string) related to the image. (FIG. 2) is further stored in the feature storage unit 21 for each candidate image in association with the image feature quantity or identification information (step S14). In this case, the text feature amount is obtained by associating, with respect to one or two or more texts related to the image, weights representing the depth of the relationship between the text and the image as illustrated in FIG. It is.

  In this way, text related to an image is a keyword such as an object, a scene, a theme, a time zone, etc. (hereinafter also referred to as a “tag”), and its typical source (acquisition source) is It is free to extract from the web page on which the image was posted, or to be associated with the image in the form of a caption or other tag description. For example, if one or more tags are given to an image on a web page, a text feature amount can be extracted in the same manner as an image feature amount by a vector space model, which is a general method for similar text search. Also, based on the keyword that is the basis for acquiring each image, the keyword itself, its superordinate concept, subordinate concept, and other related terms can be used as text feature quantities.

  By storing such a text feature amount together with the image feature amount in the feature storage unit 21, in this embodiment, it is possible to perform a similar image search in consideration of the text feature amount. However, since the source of the text feature amount is free and is not limited to extraction from the web page as described above, for example, an administrator or a user may input in advance for each image.

[3. Effect of similar image search)
Next, a flowchart of FIG. 4 shows a processing procedure for performing a similar image search using both the image feature quantity and the text feature quantity using the feature storage means 21 configured as described above. This processing procedure includes receiving a query image, searching for a similar image, and displaying a search result.

[3-1. (Reception of query image)
That is, first, the accepting unit 22 accepts a request instruction for similar image search accompanied by a query image specification from the user who uses the terminal T via the terminal T and the communication network N (step S29). Here, an example of the screen transition leading to the designation of the query image is shown. For example, as a result of inputting a desired text X1 in a character string input field F (so-called text box) and operating the search button B1 with the mouse pointer P on the screen illustrated in FIG. Assume that the result of image search via the Web image search API or the like is displayed as a list as illustrated in FIG. By using the Web image search API, comprehensive results can be obtained for text search.

  In the list display of FIG. 6, a predetermined similarity search button B2 is displayed for each of a plurality of images G hit in the search, and the accepting unit 22 queries the image corresponding to the similarity search button B2 operated with the mouse pointer P or the like. A request instruction is accepted as an image. The configuration of the user interface for accepting the designation of the image is free, and it is not essential that the search starts from the designated text as in the examples of FIGS. 5 and 6, for example, a recommended image collection or a topic image corner. It is also possible to simply accept the specification of the query image from the list of images.

  Further, the web search engine E can be used to display a list of image search results based on keyword designation as shown in FIG. 5 as shown in FIG. For example, if the keyword received on the screen as shown in FIG. 5 is transferred to the web search engine E, and a similar search button B2 is added to each image included in the search result, the screen display as shown in FIG. It is possible to reduce the processing load of the similar image search when transitioning from FIG. 5 to FIG.

[3-2. (Registering unregistered images)
When a query image is designated by an operation on the screen of FIG. 6 or the like (step S30), first, the discrimination control unit 15 first identifies the query image identification information (URL) designated by the search request received by the similar image search unit 23. Or the like), whether or not the image feature amount is stored in the feature storage means 21 (step S30), and for the image that has not yet been stored (step S30: “NO”), Image data is acquired by the image acquisition means 13, and the image feature quantity is extracted by the feature quantity extraction means 14 and then stored in the feature storage means 21 (step S31).

[3-3. (Search for similar images)
Subsequently, following the weight setting process (steps S32 to S35), which will be described later, the similar image search means 23 is designated in the request instruction by referring to the feature storage means 21 for a large number of candidate images to be searched. The similarity of the image feature amount to the query image (also referred to as “image similarity”) and the text feature amount relative to text such as a tag associated with the query image (also referred to as “reference text”) Similar images are searched based on the similarity (also referred to as “text similarity”) (step S36).

  At this time, when a query image is specified from a group of images searched based on the specified text (for example, FIGS. 5 and 6), the similar image search means 23 uses the reference text associated with the specified query image as the reference text. If at least the designated text is used and there is a tag associated with the image, the tag and the designated text are combined as a reference text. Note that when a similar image search is performed by selecting a query image without performing an image search from the designated text, only text such as a tag associated with the selected query image may be used as the reference text.

  In the search for similar images as described above, the criteria and algorithms for calculating the image similarity and the text similarity are arbitrary. For example, the image feature amounts and the text feature amounts are set as multidimensional vector data, and the distance is set. It is generally considered to calculate and compare. More specifically, for example, between the query image and each candidate image, a distance function indicating the image similarity (a function related to distance calculation in the feature amount space; hereinafter the same) and a distance function indicating the text similarity are: If the distance formula is the same, for example, Euclidean distance, a high-speed search is possible with a general multidimensional space index, but if it is not a different metric space or Euclidean distance, a metric space index represented by vp-tree. You may search using.

  The similar image search means 23 comprehensively evaluates the similarity of the image feature amount and the similarity of the text feature by addition or multiplication, so that the image similarity and the text are determined according to the setting of the calculation standard in the comprehensive evaluation. With regard to the similarity, at least one of the candidate images that is superior or the candidate image that is neither significantly inferior is prioritized as the top of the similar search results.

[3-4. Weighting in similar image search)
In addition, when calculating the similarity of a text feature amount using a plurality of texts associated with a designated query image or candidate image, the weighting unit 24 performs a predetermined weighting between the plurality of texts used for the calculation. . Here, the description will be divided into the setting of the weight for the text and the calculation of the similarity using the set weight. As for the setting of the weight, setting of the weight (for example, FIG. 2) for each candidate image and setting of the weight for the reference text side of the query image can be considered. An example is shown.

  The setting of the weight on the candidate image side is performed by determining the content of the image, the content of the web page that included the image, the content of the caption associated with the image on the web page, the XML of the web page, for example, Based on the priority order between tags associated with the image by description or the like, it is stored in advance in the feature storage means 21. The setting of the weight for the reference text may be performed when the similarity calculation is repeated for each candidate image. Here, an example is shown in which the calculation is performed prior to the calculation.

  For example, first, when the specified text exists, the specified text and the tag attached to the image searched with the specified text do not always match. In some cases, the assigned image may be searched, or some of a plurality of tags assigned to the image included in the search result may match the specified text.

  Therefore, in order to more reflect the intention of the searcher appearing in the designated text, the following weight is given in relation to the designated text. First, as illustrated in FIG. 7 (1), the tag matches the text X1 in the tag (for example, X2) given to the query image G specified from the search result based on the specified text X1 (for example, “mountain”). When there is nothing to do, the weighting means 24 adds the designated text X1 to the reference text in addition to the tag of the query image G as illustrated in FIG. For example, a predetermined weight (for example, “5”) that is relatively larger than “1”) is used.

  On the other hand, as illustrated in FIG. 8A, the text X1 is included in the tags (for example, X2 and X3) given to the query image G specified from the search result based on the specified text X1 (for example, “mountain”). If there is a match (related tag X3 “mountain”), as shown in FIG. 8 (2), the weight of the matching tag (for example, “1”) is set to a large predetermined weight as described above. (For example, “5”) is added to, for example, “6”. Note that the weight of the matching tag may be replaced with such a large predetermined weight (for example, “1” → “5”).

  In order to realize the weighting of the reference text based on the designated text as described above, in the processing procedure of FIG. 4, when there is a designated text (step S32: “YES”), if there is no designated text in the tag of the query image (step S33: “NO”), the designated text is used as the reference text in addition to the tag of the query image (step S34), and a predetermined value (“5” in the above example) is added to the weight of the designated text in the reference text. (Step S35).

  As described above, the weight set on both the text side of the candidate image and the reference text side of the query image is reflected in the calculation of the similarity in the similar image search (step S36). Candidate images with matching text will hit with high priority. For example, the similarity of the text feature amount is highest when the reference text having a large weight on the query image side matches the text having a large weight on the candidate image side.

[3-5. (Display of search results)
Then, the result display means 25 displays a search result of similar images reflecting the weights as described above (step S37). At this time, the similarity display means 26 for the searched image, the presence / absence or similarity of the image feature amount with respect to the query image and the presence / absence or similarity of the text feature amount with respect to the reference text associated with the query image And at least one of them is displayed. Here, the image similarity and the text similarity are displayed, and an example thereof is shown in FIG. This example is a display screen example of a similar image search result. The upper part of the screen is a search condition display field J, and the lower part is a display field K of similar image search results based on the condition. In the similarity display field R1 for each image of the search result, the image similarity and the text similarity are numerically displayed.

  As another example of the similarity display, if the image similarity and the text similarity are indicated by the length of the black portion of the bar graph for each image as in the similarity display column R2 shown in FIG. There is an advantage that can be grasped visually. As another example, in the similarity display column R3 shown in FIG. 11, the height of each of the image similarity and the text similarity is relatively shown by a band graph. Can be easily grasped in a simple display form.

[4. Main effect)
As described above, in the present embodiment, an image is acquired from the web server (image server) GS based on the search result obtained by using the existing web search engine E and passing the keyword to the web search engine E, and the image By storing and storing the image feature amount extracted from the image and using it for similar image search, a similar image search engine can be realized without a heavy load using a crawler independently.

  In particular, in the present embodiment, as described above, by using an image search engine that searches for an image from a keyword as the web search engine E, an image can be obtained as a search result. In comparison, the web server that provides the original image can be accessed quickly and the processing can be made more efficient.

  In the present embodiment, as described above, not only the image feature amount but also the text feature amount is stored and added to the similar image search (FIGS. 2 and 3), so that only the appearance is similar in the case of only the image feature amount. It is possible to avoid the problem that the search accuracy deteriorates due to hitting another similar image candidate.

  In addition, in the present embodiment, as described above, when the image feature amount of the image specified in the similar image search is not stored (step S30 in FIG. 4), the image is acquired and the image feature amount is extracted and stored. By performing (step S31), the search data is sequentially enriched with the operation of the similar image search, and the quality of the similar image search can be improved.

[5. Other effects
Further, in the present embodiment, as described above, when searching for a similar image based on the specified image, not only the similarity of the image feature quantity with the specified base image but also the specified base image By using the similar image narrowing based on the similarity based on the text feature amount (for example, FIG. 2) with respect to the text associated with the text (step S36 in FIG. 4), the search accuracy when the similar image search is performed using the image as a key. It is possible to prevent a decrease in the above.

  In particular, in the present embodiment described above, the specification of a query image is received from a search image based on the specified text (for example, FIG. 5), and the specified text is used as the basis of the text feature amount on the query side (see FIG. Steps S32 to S34 of FIG. 4, FIGS. 7 (1) and (2)), the text itself consciously specified can be reflected in the similar image search, so that the prevention of a decrease in search accuracy is further ensured.

  Further, in the above embodiment, when the similarity of text feature amounts is calculated using a plurality of texts associated with an image, for example, the weight of the text directly designated by the user by type input or the like is relatively set. By performing predetermined weighting such as enlarging (step S35 in FIG. 4) (FIGS. 7 (1) and (2)), a highly accurate search result suitable for the user's intention can be realized.

  Furthermore, in the above-described embodiment, the search result obtained can be easily grasped by indicating whether the obtained search result is suitable for an image or text (for example, FIGS. 9 to 11).

[6. Other embodiments]
In addition, said each embodiment is only an illustration, and this invention includes what is illustrated below and other embodiment other than that. For example, in addition to the image feature amount and text feature amount as information for each image, arbitrary information such as image data itself is stored in the feature storage unit 21 or other storage unit, and feature amount extraction based on a new standard is performed. You may utilize for an appropriate use.

  Also, the designated text used when searching for similar images is not limited to type input from a keyboard or the like, but may be designated by selecting from a list of keyword groups of interest, for example. Moreover, although the said embodiment presupposed the client server system (FIG. 1) which consists of the server 1 and the terminal T, you may implement | achieve this apparatus as a stand-alone apparatus.

  In addition, each element such as means may be realized by other information processing mechanisms such as an electronic circuit based on a wired logic or the like without being limited to an arithmetic control unit of a computer. Further, each configuration diagram, data diagram, flowchart diagram, and the like are merely examples, and the presence / absence of each element, its order, specific contents, and the like can be changed as appropriate. For example, the apparatus of the present invention may be realized by using a plurality of apparatuses such as servers, and a configuration in which each storage unit is realized by a separate and independent server apparatus or system is also common. Depending on the function, the configuration can be flexibly changed, for example, by calling an external platform or the like with an API (application program interface) or network computing (so-called cloud or the like).

DESCRIPTION OF SYMBOLS 1 Search server 6 Arithmetic control part 7 Storage device 8 Communication means 11 Index storage means 12 Search access means 13 Image acquisition means 14 Feature quantity extraction means 15 Discrimination control means 21 Feature storage means 22 Acceptance means 23 Similar image search means 24 Weighting means 25 Result display means 26 Similarity display means B1 Search button B2 Similar search button D Keyword list E Web search engine F Character string input field G Query image GS Image server J, K Condition display field N Communication network P Mouse pointers R1, R2, R3 Similarity display field T Terminal X (X1 to X3) Text X2, X3 Tag

Claims (6)

A feature information creation device that creates a feature storage unit that stores an image feature amount for each image in order to search for an image similar to a specified image,
Search access means for sending keywords to web search engines and receiving search results for the submitted keywords,
Based on the identification information of the information included in the received search results, image acquisition means for acquiring image data by accessing a web server;
Feature quantity extraction means for extracting an image feature quantity from the acquired image data and storing the image feature quantity in association with the identification information in a predetermined feature storage means;
A feature information creating apparatus characterized by comprising:   The feature information creating apparatus according to claim 1, wherein the web search engine is an image search engine that searches for an image based on a keyword.   The feature amount extraction unit further causes the feature storage unit to store a text feature amount based on the keyword transmitted by the search access unit to the web search engine in association with the image feature amount or the identification information. The feature information creation device according to claim 1, wherein the feature information creation device is a feature.   For the image specified in the search request for similar images, it is determined whether the image feature amount is stored in the feature storage means, and the image data is acquired for the image determined not to be stored. And a determination control unit that causes the image storage unit to acquire the image feature amount, to extract the image feature amount to the feature amount extraction unit, and to store the image feature amount in association with the image feature amount or the identification information in the feature storage unit. The feature information creation device according to any one of 1 to 3. A feature information creation method in which a computer creates feature storage means for storing an image feature amount for each image in order to search for an image similar to a designated image,
A search access step in which a computer sends a keyword to a web search engine and receives search results for the sent keyword;
An image acquisition step in which the computer acquires image data by accessing a web server based on identification information of information included in the received search result;
A feature amount extracting step in which a computer extracts an image feature amount from the acquired image data and stores the image feature amount in association with the identification information in a predetermined feature storage unit;
A feature information creating method characterized by comprising: A feature information creation program for causing a computer to create a feature storage unit that stores an image feature amount for each image in order to search for an image similar to a specified image,
The program controls the computer,
Have web search engines send keywords, receive search results for those keywords,
Based on the identification information of the information included in the received search results, the image data is obtained by accessing the web server,
An image feature amount is extracted from the acquired image data, and the image feature amount is stored in a predetermined feature storage unit in association with the identification information.

Images