Search Support System

The search support system addresses the limitation of name-based searches by using image processing and contextual data to recommend products that match user preferences and intended use, improving the accuracy and relevance of search results.

JP7582371B2Active Publication Date: 2024-11-13NIKON CORP
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
JP2023069459
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2009-12-24
Filing Date
2023-04-20
Publication Date
2024-11-13
Estimated Expiration
2030-12-21

AI Technical Summary

Technical Problem

Existing search systems fail to effectively assist users in finding products based on image input, especially when the user lacks product names or specific details, limiting the ability to discover similar or complementary items.

Method used

A search support system that utilizes image data processing to extract features from user input, compares them with a database, and returns similar products based on image similarity, user preferences, and contextual information such as date, location, and situation.

Benefits of technology

Enables users to find products that match their preferences and intended use by analyzing image data and providing relevant search results, enhancing the accuracy and relevance of product recommendations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007582371000001
    Figure 0007582371000001
  • Figure 0007582371000002
    Figure 0007582371000002
  • Figure 0007582371000003
    Figure 0007582371000003
Patent Text Reader

Abstract

To provide a retrieval support system capable of retrieving an object which a user wants to retrieve.SOLUTION: A retrieval support system comprises: a database in which object image data related to an object retrieved by a user is accumulated; an extraction unit for extracting a retrieval part of input image data; a retrieval unit that compares image feature information indicating a feature of an image of the image data of the retrieval part with image feature information indicating a feature of an image of the object image data in the database and that obtains from the database the object image data, which are matched or highly similar to each other from among the object image data, with respect to the image data of the retrieval part; and a condition setting unit for setting a retrieval condition including at least one of date and time information related to dates and times, place information related to places, and status information related to statuses. The retrieval unit obtains the object image data on the basis of the retrieval condition set by the condition setting unit.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a search support system. This application claims priority based on Japanese Patent Application No. 2009-292857, filed on December 24, 2009, the contents of which are incorporated herein by reference. [Background technology]

[0002] Conventionally, when a user searches for or even purchases an object (e.g., a product) via a network such as the Internet, a system has been used that assists the user in selecting or deciding on the object to be searched for (see, for example, Patent Document 1). For example, a user may use various search engines on the Internet to input the name of an object (e.g., a product name, etc.) or characteristics of the object to search for the corresponding object. If necessary, the user may also search for stores that sell the above-mentioned corresponding object. A server device having a search engine such as that described above assists the user in searching for objects by presenting images or prices of objects that correspond to the user's preferences at the searched stores via the Internet to the user's terminal, or by providing related objects.

[0003] Furthermore, in recent years, when purchasing goods, people often purchase goods from virtual stores in virtual malls on the Internet, rather than visiting a retail store that sells goods face-to-face. As a result, even in areas where there are no large retail stores nearby, it is now possible to purchase a favorite product from a wide range of products without leaving the comfort of one's own home (see, for example, Patent Document 2). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2004-246585 A [Patent Document 2] JP 2002-150138 A Summary of the Invention

[0005] An aspect of the present invention includes a database in which object image data relating to an object searched for by a user is stored, an extraction unit that extracts a search portion of input image data, and a search unit that compares image feature information indicating features of an image of the image data of the search portion with image feature information indicating features of an image of the object image data in the database, and retrieves from the database object image data that matches or is highly similar to the image data of the search portion. Based on the input image data, Search criteria that include at least one of the following: date and time information, location information, and situation information. Extract the search conditions. and a condition setting unit that sets the search conditions, and the search unit acquires the object image data based on the search conditions set by the condition setting unit. [Brief description of the drawings]

[0006] [Figure 1] FIG. 1 is a block diagram showing an example of the configuration of a search support system. [Figure 2A] 2 is a conceptual diagram showing an example of the configuration of a product table stored in the database 16 of FIG. 1. FIG. [Figure 2B] 2 is a conceptual diagram showing an example of the configuration of a product table stored in the database 16 of FIG. 1. FIG. [Diagram 3] 2 is a conceptual diagram showing an example of the configuration of a user registration table stored in the database 16 of FIG. 1. [Figure 4] 11 is a flowchart showing an example of a product search operation in the search support system. [Diagram 5] 10 is a flowchart showing an example of an operation for collecting preference information in the search support system. [Figure 6] FIG. 13 is a block diagram showing a modified example of the configuration example of the search support system. [Figure 7]FIG. 1 is a block diagram showing an example of the configuration of a search support system. [Figure 8] 8 is a table recording attribute information of combined product items in category 1 (shirts) of product item database 26 in FIG. 7. [Figure 9] 8 is a table recording attribute information of combination product items in category 2 (trousers) in the product item database 26 in FIG. 7. [Figure 10] 8 is a table recording attribute information of combined product items in category 3 (jackets) in the product item database 26 in FIG. 7. [Figure 11] This is a table showing the correspondence between the sales product items in category 1 being sold and the combination product items in category 1 similar to these collected from image media. [Figure 12] This is a table showing the correspondence between the sales product items in category 2 being sold and the combination product items in category 2 similar to these collected from image media. [Figure 13] This is a table showing the correspondence between the sales product items in category 3 that are being sold and the combination product items in category 3 that are similar to these and that have been collected from image media. [Figure 14] 8 is a table showing combinations between each category of combined product items collected from image media in combination information database 28 in FIG. 7. [Figure 15] 8 is a table recording attribute information of combined product items in category 1 (shirts) of combined product image information database 29 in FIG. 7. [Figure 16] 8 is a table recording attribute information of combined product items in category 2 (trousers) of combined product image information database 29 in FIG. 7. [Figure 17] 8 is a table recording attribute information of combined product items in category 3 (jackets) of combined product image information database 29 in FIG. 7. [Figure 18] 8 is a flowchart showing an example of the operation of the search support system in FIG. 7. [Figure 19] 8 is a conceptual diagram showing the configuration of a user table in the history database 31 of FIG. 7. [Figure 20] 8 is a conceptual diagram showing the configuration of a purchase history table in the history database 31 of FIG. 7. [Figure 21] FIG. 1 is a conceptual diagram illustrating segmentation of image data by the Graph-Cut method. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0007] In the conventional examples described in Patent Documents 1 and 2 mentioned above, when a user searches for an object, the user inputs text data indicating the name of the object, or selects from images of the object prepared in advance for searching. Therefore, it has been found that if the user only has an image of the object and does not know the name of the object, for example, there is a problem that the user is unable to search for the object that he or she actually wants to search for, or an object similar to the object that he or she is searching for. Therefore, aspects of the present invention provide a search support system, a search support method, and a search support program that can search for an object that a user wishes to search for. <First embodiment> A search support system according to an embodiment of the present invention will now be described with reference to the drawings, in which: Fig. 1 is a block diagram showing an example of the configuration of the search support system according to this embodiment. In this figure, the search support system has a search server 1 and a user terminal 2 connected to the search server 1 via an information and communication network I such as the Internet. Here, the user terminal 2 is a terminal owned by a user, and is identified by user identification information unique to each user. In the following, in this embodiment, the search and purchase of products such as fashion items, clothes, shoes, necklaces, and hats will be described as examples of objects that a user may wish to search for. Note that this embodiment is not limited to products such as fashion items, clothes, shoes, necklaces, and hats, and can also be applied to objects that can be searched for via the Internet or a network, such as electrical appliances, furniture, and paintings. Examples of objects in this embodiment include products, articles, electrical appliances, furniture, paintings, buildings including stores and companies, plants, animals, real estate (including the exterior, interior, and layout of an apartment building), landscapes, and the like.

[0008] The search server 1 is a server that assists the user in searching for products in stores on the Internet, and has an extraction unit 11, a type determination unit 12, a search unit 13, a combination extraction unit 14, a preference extraction unit 15, a database 16, a transmission / reception unit 17, and a condition setting unit 18. As shown in FIG. 2A, the database 16 stores a product table in which the name of the product, product image data (object image data) which is image data of the product, feature information extracted from the product image data, store information selling the product, the price of the product, combination information with other products to be combined with the product, type identification information indicating the type of the product (product genre such as clothes, shoes, hats, necklaces, etc.), and use information as attribute information indicating the use purpose are associated with each other, in association with product identification information identifying each product. For example, product information from stores registered as members in the search support system is sequentially stored in the database 16. The feature information is numerical data (for example, a vector of a dimension corresponding to the number of elements) of elements such as color, shape, pattern, and texture. In addition, when the image of the product that the user wants to search for is directly compared with the product image (object image) stored in the database 16, the type determination unit 12 may be omitted.

[0009] In addition, as the usage information, time information, place information, and occasion information are set. As the time information, information related to time such as date, time, morning, mid-morning, afternoon, night, season (spring, summer, fall, winter, all year round), national holidays, Children's Day, Doll's Day, and other festivals are set. As the place information, information related to places such as domestic, overseas (place names in Asia, Europe, and the United States), parks, resorts, cities, the sea, mountains, plateaus, and high-end restaurants are set. As the situation information, information such as commuting, going to school, part-time work, going to a resort, going to a city, welcome parties, farewell parties, domestic travel, overseas travel, driving, hiking, cycling, walking, year-end parties, New Year's parties, joint parties, matchmaking, divorce meetings, funerals, kindergarten entrance ceremonies, kindergarten graduation ceremonies, entrance ceremonies (for oneself or children), graduation ceremonies (for oneself or children), mountain climbing, swimming in the sea, going to plateaus, wedding receptions, company entrance parties, graduation parties, wedding anniversaries, birthdays, barbecues, and high-end restaurants are set. The situation information may also include the type of clothing that the user wishes to wear depending on the situation, such as traditional, feminine, casual, sporty, dressy, career, formal, elegant, sweet, cool, casual Friday, cool biz, warm biz, character, stylish, etc. Furthermore, as the situation information, weather information indicating the weather, for example, sunny, cloudy, rainy, snowy, typhoon, storm, etc. may be used. Furthermore, as the situation information, psychological information regarding the deep psychology, for example, lack of energy, melancholy, dark, anxious, sad, fun (feeling excited), happy, gentle, calm, intense, etc. may be used. Furthermore, as the situation information, influence information regarding the impression given to the surroundings, for example, plain, dark, bright, gentle, elegant, clean, healthy, feminine, masculine, intelligent, friendly, refreshing, etc. may be used. Fig. 2B shows an example of usage information (date and time information, place information, situation information) corresponding to a product.

[0010] The condition setting unit 18 may be configured to divide the date input as date and time information into a range of spring, summer, autumn, and winter preset inside by month and day, and change the input date into a season for use. For example, spring is divided into March 15th to June 14th, summer into June 15th to September 14th, autumn into September 15th to December 14th, and winter into December 15th to March 14th, and when August 10th is input as date and time information, the condition setting unit 18 determines that it is the summer season and uses this season as purpose information. The condition setting unit 18 may be configured to divide the time range into early morning, morning, afternoon, and night, and change the input time into a time period for use. For example, early morning is divided into 4:00 AM to 6:59 AM, morning into 7:00 AM to 11:59 AM, afternoon into 12:00 PM to 4:59 PM, and night into 5:00 PM to 3:59 AM, and when 10:00 AM is input as date and time information, the condition setting unit 18 determines that it is a morning time period and uses this time period as purpose information. The above-mentioned use information is displayed on the display screen of the user terminal 2 as a combo box (for example, a combination of a rectangular area (text box) for character input and an item selection list (list box)), and is selected from this combo box with a pointing device. For example, when an item is selected by clicking with a mouse, the user terminal 2 sets the item of use information in the combo box selected by clicking as the use information selected by the user. The image information of the combo box displayed on the display screen of the user terminal 2 may be installed in advance as an application program in the user terminal 2, or the user terminal 2 may be configured to read the image information of the combo box from the search server 1 when the search processing application is started.

[0011] The extraction unit 11 receives input image data (image data provided by the user) sent by the user from the user terminal 2 via the transmission / reception unit 17, and extracts the search portion from the input image data. More specifically, the extraction unit 11 extracts the contour (search portion) of the image area of ​​the product from the input image data, and generates contour image data (image data of the search portion). Here, in the process of extracting contour image data, a contour is a portion where the density value of an image changes suddenly, and a differential operation is performed to extract this. However, since data is arranged at regular intervals in a digital image, the differential is approximated by an operation (difference) that takes the difference between adjacent pixels, and the portion where the density of pixels changes suddenly is extracted as the contour.

[0012] From the outline image data, the type determination unit 12 searches for template image data corresponding to the outline image data from template image data in an outline data table of products previously stored in the database 16, and reads type identification information corresponding to the template image data set in the database 16. It is preferable that this template image data is stored in advance in the outline data table of the database 16 as templates to be compared with the outlines of a large number of products of each type photographed from a plurality of angles.

[0013] The search unit 13 extracts features of the search image data obtained by extracting the contour of the input image data in order to extract similar image data from the storage area of ​​the type corresponding to the type identification information. The search unit 13 searches image data having type identification information corresponding to the determined type of the product table in the database 16 for feature information that matches or is highly similar to the feature information obtained by the feature extraction, and extracts a preset number of image data of products with high similarity in order of similarity. Here, when the type determination unit 12 is not provided, the search unit 13 compares the image of the product obtained by extracting the contour of the input image data with the product image data in the entire product table, and extracts product image data that is highly similar to the image of the product obtained by extracting the contour. Furthermore, the condition setting unit 18 reads in the purpose information transmitted by the user from the user terminal 2 together with the input image data via the transmission / reception unit 17 , sets this purpose information internally, and also outputs it to the search unit 13 . When extracting product image data in the order of the degree of similarity described above, the search unit 13 extracts product image data in the order of the degree of similarity from among the image data having the usage information input from the condition setting unit 18. Therefore, the search unit 13 searches the product table for products that are similar to the input image data and have the usage information set by the user. As a result, the search unit 13 can extract products that have a shape that the user likes and that correspond to the user's usage from the product table. Furthermore, the search unit 13 reads out product information (object information) corresponding to the searched product image data from the product table, and transmits it to the user terminal 2 of the user. Note that the above-mentioned product information is information including at least one of the name (product name), a store selling the product corresponding to the product image data, the URL of the store, the telephone number or address of the store, and the price of the product.

[0014] The preference extraction unit 15 randomly extracts a preset number of image data of other products (products of the same type as the product searched for by the user) included in the type of product searched for by the user from the product table, and transmits the image data to the user terminal 2. The user terminal 2 displays the image data transmitted from the search server 1 on a display screen (not shown). Furthermore, the preference extraction unit 15 inputs the evaluation results (e.g., judgment results of whether the user likes or dislikes the product) for each product (e.g., product image data stored in the database 16) sent from the user terminal 2, and writes and records preference information including at least one of shape, color, pattern, and texture for each user in the user registration table shown in Figure 3 in the database 16.

[0015] Here, the user registration table stores purchase information (including purchased product identification information, purchase date and time, purchase store, purchase price, etc.) corresponding to the user identification information, which indicates what products the user has purchased in the past, preference information extracted by the preference extraction unit 15, and point information indicating the number of points the user has. These points can be used in the same way as cash when paying for the purchase of products. That is, when discount coupon information is added to the above-mentioned store information and the user purchases the searched product using the discount coupon information, the search server 1 stores the purchase information on the product in the user registration table of the database 16. Then, the search server 1 adds points corresponding to the price of the purchased product as the number of points in the point information to the target user in the user registration table. Note that the preference extraction unit 15 may extract preference information for each type from the purchase information. The combination extraction unit 14 searches for other types of products that are likely to be purchased in combination with the extracted product, and transmits the products to the user's terminal, the products corresponding to combination information that is stored in advance in correspondence with the preference information.

[0016] Next, the operation of the search support system in this embodiment will be described with reference to Fig. 1 and Fig. 4. Fig. 4 is a flowchart for explaining an example of the operation of the search support system in this embodiment. In the following explanation, a case where the type determination unit 12 is provided will be described. When the type determination unit 12 is not provided, the processing of the type determination unit 12 is omitted from the flowchart. A user reads an image of a model wearing clothes that the user likes from a magazine or the like using a color scanner or the like connected to the user terminal 2, or takes a picture of the model using a digital camera, or downloads the image from the Internet (step S1). Then, the user inputs the scanned, photographed, or downloaded image into the user terminal 2 as input image data, and determines whether or not to select an area of ​​the image of the product to be searched for depending on whether or not multiple types of products are included in the input image data (step S2). Note that the above input image data may be stored in advance in the user terminal 2, or may be stored in an external terminal.

[0017] At this time, for example, when searching for the product name and stores selling a skirt, if the input image data shows only a skirt, the process proceeds to step S4. On the other hand, if the entire model has been photographed, and not just an image of the skirt portion, the process proceeds to step S3. Then, the user selects the input image data by marking the image of the skirt area with a line or the like using an image processing tool installed on the user terminal 2 (any tool that can process the input image data by drawing lines) (step S3), and then proceeds to step 4. At this time, the user also inputs to the user terminal 2 purpose information to be used when searching for the input image data, for example, at least one of date and time information, location information, and situation information.

[0018] The user terminal 2 transmits the input image data entered by the user (including the data from which the image was selected) and the purpose information together with its own user identification information as a search request signal to the search server 1 via the information and communication network I (step S4).

[0019] In step S2 described above, the user terminal 2 temporarily transmits input image data (an image of the entire model) to the search server 1. The search server 1 determines whether or not the input image data received from the user terminal 2 includes multiple types of products, and extracts the determined multiple types of products as images. The search server 1 may then transmit the extracted multiple types of image data to the user terminal 2. In this case, in step S3, the user selects image data of the type of product that the user wishes to search for from the multiple types of image data extracted from the input image data received from the search server 1, and proceeds to step S4.

[0020] Next, when the transmitting / receiving unit 17 in the search server 1 receives the search request signal from the user terminal 2, it outputs the input image data to the extraction unit 11 and outputs the registration information to the condition setting unit . The extraction unit 11 extracts the contour of the product image in the input image data input from the transmission / reception unit 17 (extracts the search portion). At this time, if none of the areas in the input image data are marked, the extraction unit 11 extracts the contour image data of the entire input image data, whereas if there is a marked area, the extraction unit 11 extracts the contour image data of the image of that area. Then, the type determination unit 12 searches for template image data corresponding to the outline image data from an outline data table corresponding to the product type stored in advance in the database 16, and reads type identification information corresponding to the template image data from the outline data table (step S5). At this time, the search unit 13 resets to 0 the number of products counted as the extraction result to be transmitted to the user.

[0021] The search unit 13 extracts features of image data (search image data) to be searched for in the input image data, corresponding to the inside of the contour in the above-mentioned contour image data, and searches the product table in the database 16 for product image data similar to the feature information of the search image data (e.g., information based on color and shape) (step S6). At this time, the search unit 13 obtains the vector distance for each element of the feature information (for example, if the feature information is color-based information, then blue, red, yellow, etc.), and calculates the similarity by whether this distance matches or is close or far. Note that a product with a close distance is considered to have a high degree of similarity, and the product with the highest degree of similarity is extracted. Here, products that have already been extracted as having the highest degree of similarity and stored as an internal compared file are excluded, and the product image data with the highest degree of similarity is extracted from among the products remaining in the product table.

[0022] Next, the search unit 13 compares the use information set in the condition setting unit 18 with the use information of the product corresponding to the extracted product image data, and determines whether or not they match (step S7). At this time, the search unit 13, for example, compares multiple items (date and time information, location information, situation information) of each of the usage information with each item of the usage information set in the condition setting unit 18, and if there is a match in at least one of the items, it determines that the usage information of the extracted product matches the usage information registered in the condition setting unit 18. Furthermore, the search unit 13 writes the compared products, i.e., the products extracted as having a high similarity, into an internal compared file, and proceeds to step S9 (step S8). Then, the search unit 13 stores (writes) the product identification information of the product whose usage information matches in an internal extraction file as the extraction result, and increments the extraction number indicating the number of products in the extraction result by 1. On the other hand, if the usage information does not match, the search unit 13 does not perform the process of writing the product identification information of this product in the extraction file or the process of incrementing the extraction number by 1, and proceeds to step S9 (step S7).

[0023] The search unit 13 determines whether the extraction number matches a preset number of products to be sent to the user (step S9), and if the extraction number matches the number of products to be sent to the user, proceeds to step S10; on the other hand, if the extraction number does not match the number of products to be sent to the user, proceeds to step S6.

[0024] Next, the search unit 13 reads out product identification information of the searched product from the extraction file, reads out product information of the product corresponding to this product identification information (information including at least one of the product name, product price, product image data sales store, URL, store telephone number and address) from the product table, and transmits it to the user terminal 2 of the user via the transmission / reception unit 17 (step S10). Here, as an example, the transmission / reception unit 17 reads out the network address of the user terminal 2 that is added when a search request signal is transmitted from the user terminal 2, and transmits the above product information to this address. Then, the user terminal 2 displays the information on the product name, image data, store and price of the search results transmitted from the product sales search server 1 on the display screen. In the above-mentioned step S5, the search server 1 may extract the product image data to be searched from the database 16 based on the input image data included in the search request signal only when it is detected that the user identification information included in the search request signal is registered in the user registration table in the database 16. In addition, when the search server 1 detects that the user identification information included in the search request signal is not registered in the user registration table, it may transmit to the user terminal 2, for example, a message indicating that user registration based on the user identification information is required.

[0025] In this way, by sending image data of a favorite product and usage information, which is information on the purpose of this product, from the user terminal 2 to the search server 1, the user can obtain information such as the product name of the favorite product or a product similar to the favorite product, store information indicating the store that sells the product, and the price of the product as extraction results from the search server 1.

[0026] In the above-described embodiment, the user has added use information to search request information and transmitted it from the user terminal 2. However, the condition setting unit 18 may be configured to determine the material (cotton, wool, silk, cashmere, etc.) of the product to be searched for from the input image data, and determine the season (spring, summer, autumn, winter) as date and time information from this material, or extract conditions such as situation information from the color, etc., and use these as search conditions. The condition setting unit 18 may also be configured to extract a person from the input image data, and extract search conditions from the background image of the person to use as the search conditions. In this case, for example, the condition setting unit 18 has a database for extracting usage data, identifies a tree in the background image, searches for a similar image from this database for extracting usage data, and can determine the season from the state of the flowers and leaves, or determine location information from character data in the background image.

[0027] In the above-described embodiment, the use information is added as attribute information of each product in the product table, but a product table may be configured for each category of use information. For example, product tables for spring, summer, autumn, and winter may be configured and used. In this case, the search unit 13 selects a product table corresponding to the registration information added to the search request information from the user terminal 2, and searches for products similar to the input image data from the selected product table, thereby shortening the search time.

[0028] Here, discount coupon information is added to the store information, and when the searched product is purchased using this discount coupon information, the product sales search server 1 charges the target store an amount corresponding to the price of the product.

[0029] Next, a process in which the search server 1 according to the present embodiment extracts user preference information and presents other products that can be combined with the searched product will be described with reference to Figures 1 and 5. Figure 5 is a flow chart for explaining an example of an operation in which user preference information is collected and other products are presented based on the preference information. The preference extraction unit 15 extracts a preset number of shoes from the product table if the type of the product is the same as that of the product searched for by the user, for example, if the product searched for by the user is shoes (step S11), and transmits the product image data and product identification information to the user terminal 2 (step S12). At this time, the products of each type that the search server 1 transmits to the user are characteristic products that differ in color, shape, pattern, etc., and are set as products from which a fashion coordinator can extract user preference information.

[0030] Then, when the user terminal 2 receives the product image data of the products for preference extraction from the search server 1, the user terminal 2 sequentially displays the product image data (display image) of the products on the display screen. At this time, if the user likes the product in the displayed product image data in the display image displayed on the display screen, the user clicks the "Good" button using an input means such as a mouse, whereas if the user does not like the product in the displayed product image data, the user clicks and selects the "No" button using the mouse or the like (step S13). When the "Good" or "No" button is selected, the user terminal 2 displays the product image data of the next product on the display screen, and this process continues until the user has finished selecting whether to accept or reject all or a predetermined number of the received product image data.

[0031] When the quality judgment for the above-mentioned product image data is completed, the user terminal 2 transmits the judgment result information, which corresponds to the quality judgment data for each product identification information, to the search server 1 together with its own user identification information (step S14). When the judgment result information is input, the transmission / reception unit 17 determines whether the added user identification information is registered in the user registration table, and if registered, outputs the above judgment result information to the preference extraction unit 15.

[0032] Next, when the judgment result information is input, the preference extraction unit 15 selects preference information corresponding to the pass / fail judgment pattern from the preference information table in the database 16. Here, the pass / fail judgment pattern refers to a data pattern of pass / fail judgments corresponding to the product identification information arranged in a predetermined order. Then, the preference extraction unit 15 writes and stores the obtained preference information in the user registration table in the database 16 in association with the user identification information (step S15).

[0033] Next, the combination extraction unit 14 selects product combination information corresponding to the above-mentioned preference information from the combination information table in the database 16. For example, when the combination extraction unit 14 extracts preference information for shoes from the combination information table, the combination extraction unit 14 selects combination information corresponding to the preference for clothes, bags, hats, etc. This combination information table stores in advance combinations of elements (shoes, clothes, bags, hats, etc.) in clothing categories corresponding to preferences. Then, the combination extraction unit 14 extracts product image data such as clothes, bags, hats, etc. that match or are similar to the combination information selected from the combination information table (extraction process for searching for similar products using the feature information (combination preference information) already described) (step S16), and transmits wearing image data of the model wearing the selected clothes, bag, hat, etc. to match the shoes to the user terminal 2 (step S17), adding combination identification information for each combination to a single type if there is only one combination, or multiple types if there are multiple combinations.

[0034] Upon receiving the wearing image data, the user terminal 2 displays the wearing image data on the display screen. This allows the user to obtain information on fashion combinations that match the user's preferences for the products that the user has already searched for or purchased. Here, when the user performs the pass / fail judgment as described above, the user terminal 2 associates pass / fail judgment data with each combination identification information, adds its own user-specific identification information as judgment result information, and transmits it to the search server 1.

[0035] Then, when the judgment result information is input, the transmission / reception unit 17 determines whether the added user identification information is registered in the user registration table, and if registered, outputs the above judgment result information to the preference extraction unit 15. The preference extraction unit 15 generates combination preference information based on the pass / fail judgment pattern corresponding to the input combination identification information. This combination preference information corresponds to the shape, color, pattern, texture, etc. of the shoes, and is made up of combinations of shapes, colors, patterns, and textures of other clothes, bags, hats, etc. that have been judged to be good, and is learned every time the user purchases a product, improving the accuracy with which the combination preference information used to extract products similar to each product included in the combination information matches the user's preferences. In other words, if a user then purchases a bag using this combined wearing image data, preference information for the bag will be searched for, and combination preference information for shoes, clothes, hats, etc. corresponding to this preference information will be obtained, thereby sequentially narrowing down the preferences of each user.

[0036] In addition, as a challenge combination (a combination that a user would not normally select, or a combination different from the above-mentioned preference information), several products corresponding to combination information that the fashion coordinator considers to be the polar opposite of the combination information selected from the preference information may be randomly selected from the product table, and these selected products may be inserted into some of the combinations when generating the wearing image data. This may lead to more diversified user preferences and increased purchasing desires.

[0037] Also, as shown in FIG. 6, the search support system in this embodiment may include a user terminal 2a configured to have an extraction unit 11a having the same extraction function as the extraction unit 11 described above, and a search server 1a excluding the extraction unit 11. In this case, the extraction unit 11a may extract a search portion (e.g., a product portion) of input image data acquired by the user terminal 2a or input image data stored in the user terminal 2a based on the input image data, and transmit only image data corresponding to the search portion as input image data to the search server 1a. Furthermore, since the components other than the extraction unit 11a are the same as those in the above-mentioned embodiment, their explanations are omitted. In this way, the user only needs to transmit image data corresponding to the search portion of the input image data to the search server 1a as image data to be transmitted, and there is no need to transmit image data corresponding to the user's face or other people's face to the search server 1a, for example.

[0038] The search support system in this embodiment can be applied not only to products such as clothes and shoes, but also to the above-mentioned objects. For example, the search support system in this embodiment can be applied to services provided in accommodation facilities (hotels, inns, etc.) and restaurants. For example, in the above-mentioned accommodation facilities, the search support system in this embodiment can be applied when selecting a combination of room layouts, a combination of window views, a combination of a room or its layout and a window view, etc. For example, in the above-mentioned restaurants, the search support system in this embodiment can be applied when selecting a combination of restaurant interiors, a combination of window views, a combination of music, a combination of interiors and window views, a combination of interiors and music, a combination of window views and music, etc.

[0039] <Second embodiment> A search support system according to an embodiment of the present invention will now be described with reference to the drawings. Conventionally, when purchasing a combination product that combines multiple different types of products, or when selecting from the vast range of products in the virtual store, users have selected and purchased products based on information they have collected themselves (accessed via the Internet, etc.). For example, when purchasing a combination product that combines multiple products of different types (categories) (e.g., clothing, combined furniture, combined home appliances, etc.), unless you actually go to a retail store, you cannot directly ask a store clerk which combination is appropriate. In addition, when selecting a product that matches well with a product that the user has purchased or plans to purchase from the vast product group in the virtual store, the user is unable to check with a coordinator of combined products one by one, and is unable to make the selection by himself / herself, resulting in the user being unable to make the purchase.

[0040] As an example of the above, when it comes to clothing, when a user wants to choose pants or a shirt to go with the jacket they are purchasing, the user makes the selection and purchase based on information obtained from magazines or advisors (salespeople, experts, acquaintances). For example, in the case of clothing, even if a user wants to choose pants or a shirt that go with a jacket he or she is purchasing, he or she cannot check with a top fashion coordinator, and is worried about selecting an impossible combination, which may lead to the user giving up on purchasing the matching pants or shirt. Here, clothing refers to all the clothes and accessories (decorations, bags, shoes, hats, etc.) that a person wears on their natural body.

[0041] In this embodiment, when a user searches for an object, the search can be supported by providing information or the like. For example, when purchasing a combined product, products that match the product that the user is purchasing or has purchased can be extracted from a group of products, and the extracted products can be recommended to the user, thereby assisting the user in purchasing the combined product.

[0042] 7 is a block diagram showing an example of the configuration of a search support system according to this embodiment. This embodiment is a system that supports a user in purchasing a sales item of any category in a product item combination consisting of sales product items of multiple categories by presenting sales product items of the remaining categories in the product item combination that are considered to be a good combination with the purchased sales product item as recommended products, thereby allowing the user to easily select a sales product item that is considered to be a good combination with the sales product item to be purchased from many sales product items. In the following description, each product in clothing (products in categories such as blazers, shirts, and pants) in general is referred to as a product item, a product item actually sold by a virtual shop is referred to as a sales product item, and each product item combined in an image medium described later is referred to as a combination product item.

[0043] In this figure, the search support system is composed of a product search support device 100 and one or more user terminals 200. The product search support device 100 has a control unit 21, a similar item search unit 22, a combination search unit 23, a product search unit 24, a transmission / reception unit 25, a product item database 26, a product item image database 27, a combination information database 28, a combined product image information database 29, a combined product image database 30, and a condition setting unit 32. The user terminal 200 is a personal computer equipped with an Internet browser, which is installed in each user's home, for example. The product search support device 100 and the user terminal 200 are connected to each other via an information and communication network I including the Internet.

[0044] The product item database 26 has a storage area configured as a plurality of tables for each type of product item. That is, as shown in FIG. 8, the product item database 26 has a storage area in a table format in which attribute information of the product items being sold is stored for each of the identification information A1 to An of the product items of shirts in category 1. Similarly, the product item database 26 has a storage area in a table format in which attribute information of the corresponding sales product item is stored for each of the identification information B1 to Bn of the product items of trousers in category 2, as shown in FIG. Similarly, the product item database 26 has a storage area in a table format in which attribute information of the corresponding sales product item is stored for each of the identification information C1 to Cn of the jacket product items in category 3, as shown in FIG. Furthermore, although not shown, the product item database 26 has a storage area that stores a plurality of tables for other clothing categories such as shoes, blouses, bags, etc., in addition to the above-mentioned categories. In addition, each table of the product item database 26 has a column for similar product items as one piece of attribute information, and product items in the combined product image information database 29 that are identical or similar to the sales product item are stored in correspondence with each other as groups of similar product items for each sales product item, as shown in Figure 11 (category 1-shirts), Figure 12 (category 2-trousers), and Figure 13 (category 3-jackets).

[0045] Here, in the product item database 26 (category 1 to category 3), the attribute information includes identification information of combined product items (product items in the combined product image information database 29 described later) associated as identical or similar, image feature data extracted from image data of the sold product item, sales information such as the price and brand of the sold product item, and the use information described in the first embodiment. The types of image feature data and the method of acquiring them will be described later. The product item image database 27 stores image data for each sales product item of each category stored in the product item database 26 in correspondence with the identification information of each sales product item.

[0046] The combination information database 28 stores combinations of clothing items worn by models, etc. in fashion magazines, fashion catalogs, and image media on the Internet, that is, each of the combined product items worn in combination, as combined product items, and stores combinations of the combined product items and the corresponding combinations of the combined product items and the combined product items in correspondence with each other, with their respective identification numbers corresponding to each other. For example, in a fashion magazine, if a model wears product items of a shirt with identification information a1-1, trousers with identification information b1-2, and a jacket with identification information c1-7, the combined product items of the identification information a1-1, the identification information b1-2, and the identification information c1-7 are stored in correspondence with each other as a set, as shown in the first row of the combination information database 28 in Fig. 14. This combination information is a combination of product items that a fashion designer has created as a combination, or a combination of product items that a fashion coordinator has created as a fashion designer. Therefore, each combination of product items is a sophisticated combination, and there is a high possibility that people who see the fashion will naturally accept it without feeling uncomfortable when wearing it.

[0047] The combined product image information database 29 is a database that accumulates, by category, combined clothing product items obtained from image media (materials or devices that present images such as magazine photos and illustrations) such as fashion magazines, fashion catalogs, and designer collection information on the Internet. Here, the combined product image information database 29 is configured, for example, in the table configurations shown in Figs. 15, 16, and 17. Fig. 15 shows a combination product item of a shirt stored as category 1, Fig. 16 shows a combination product item of trousers stored as category 2, and Fig. 17 shows a combination product item of a jacket stored as category 3. In the combined product image information database 29, identification information is assigned to the combined product items, and attribute information of the corresponding combined product items is stored for each identification information. This attribute information includes, for example, identification information of similar sales product items, feature data of images extracted from image data of the combined product items, sales information such as the price and brand of the combined product items, and use information described in the first embodiment. The types of feature data and the method of acquiring them will be described later. In addition, each table of the combined product image information database 29 has a column for similar product items as one piece of attribute information, and product items in the product item database 26 that are identical to or similar to the combined product items are stored in correspondence with each other as shown in Figure 15 (category 1-shirts), Figure 16 (category 2-trousers), and Figure 17 (category 3-jackets).

[0048] In the combined product image information database 29, in the table of Figure 15, identification information a1-1 to a1-q are stored as combined product items that are the same as or similar to the sold product item A1, identification information a2-1 to a2-k are stored as combined product items that are the same as or similar to the sold product item A2, and identification information an-1 to an-m are stored in order as combined product items that are the same as or similar to the sold product item An. Similarly, in the combined product image information database 29, in the table of Figure 16, identification information b1-1 to b1-m are stored as combined product items that are the same as or similar to the sales product item B1, identification information b2-1 to b2-r are stored as combined product items that are the same as or similar to the sales product item B2, and in turn, identification information bn-1 to bn-q are stored as combined product items that are the same as or similar to the sales product item Bn. Similarly, in the combined product image information database 29, in the table of Figure 17, identification information c1-1 to c1-q are stored as combined product items that are the same as or similar to the sales product item C1, identification information c2-1 to c2-k are stored as combined product items that are the same as or similar to the sales product item B2, and in turn, identification information cn-1 to cn-m are stored as combined product items that are the same as or similar to the sales product item Cn. The combined product image database 30 stores image data for each combined product item of each category stored in the combined product image information database 29 in correspondence with the identification information of each combined product item.

[0049] In addition, the above-mentioned product item database 26, product item image database 27, combination information database 28, combined product image information database 29 and combined product image database 30 may be classified by gender, age group, fashion trend (for example, casual, conservative, older brother style, mature style, street style, Ura-Hara (Harajuku) style, and mode style for men, while gal style, older sister style, conservative style, teen style, celebrity style, etc. for women), etc. When categorizing in this way, before processing the coordination, it is necessary to obtain from the user information on the classification, such as gender, age group (teens, twenties, thirties, etc.), fashion trends, etc.

[0050] Returning to Fig. 7, the control unit 21 is started when the virtual shop it manages is accessed from the user terminal 200, and transmits to the user terminal 200 a processing program (operated by the browser of the user terminal or the like) that acquires classification information such as gender, age, and fashion trends, and displays and selects combined items (described later). Here, the control unit 21 and the user terminal 200 transmit and receive various data via the information and communication network I and the transmitting / receiving unit 25. The user terminal 200 starts the processing program in an internal browser, and transmits and receives data to and from the product search support device 100 in the process of displaying images or selecting sales product items or combined product items. In addition, when the control unit 21 receives response data to the above screen data from the user terminal 200, it selects one of the product item tables from multiple product item databases 26 corresponding to the gender, age, and fashion trend, and transmits category information indicating the category of the combined product item, such as text information such as shirts, trousers, jackets, shoes, or image information, to the user terminal 200 via the transmission / reception unit 25 and the information communication network I.

[0051] In addition, when the control unit 21 receives category information transmitted from the user terminal 200 indicating the type of category selected by the user, for example, shirts, it selects the table shown in FIG. 8 in which shirts (category 1) is the category in the product item database 26, reads the product item identification information A1 to An in this table, reads out the image data of the product items with identification numbers A1 to An from the product item image database 27, converts them into image data of thumbnail images (reducing the number of bits of the image data), adds the identification numbers of the corresponding product items for sale to each of them, and transmits them to the user terminal 200. Furthermore, when the control unit 21 inputs the identification number of the sales item that the user selected to purchase from the thumbnail image, it processes the order and transmits the input identification number to the similar item search unit 22. The order processing here includes billing the user terminal 200 for the fee corresponding to the price stored as attribute information, checking the stock of the ordered sales item, and carrying out procedures for shipping the sales item to the address input by the user.

[0052] The condition setting unit 32 reads the identification number of the thumbnail image selected by the user and the use information transmitted from the user terminal 200 via the transmission / reception unit 25, sets this use information internally, and outputs it to the similar item search unit 22. The configuration and operation of this condition setting unit 32 are similar to those of the condition setting unit 18 in the first embodiment. The similar item search unit 22 searches the table (category 1) of the combined product image information database 29 for a plurality of combined product items in which an identification number matching the identification number of the input sales product item is written (set) in the similar product item column, extracts combined product items that are the same as or similar to the above sales product item as similar product items, and outputs them to the combination search unit 23. At this time, the similar item search unit 22 selects from the table (category 1) of the combined product image information database 29 a plurality of combined product items in which an identification number matching the identification number of the input sales product item is written in the similar product item column, and further selects product items having the same usage information as the usage information set in the condition setting unit 32, and extracts them as similar product items. That is, the similar item search unit 22 compares the use information set in the condition setting unit 32 with the use information of the product item in the combined product image information database 29, and determines whether they match. At this time, the similar item search unit 22 compares, for example, multiple items (date and time information, location information, situation information) of each use information with each item of the use information set in the condition setting unit 32, and if there is at least one match in each item, determines that the use information of the extracted product matches the use information registered in the condition setting unit 18.

[0053] The combination search unit 23 searches the combination information database 28 for product items of a different category, for example, trousers (category 2), that correspond to each of the similar product items having the identification numbers input from the similar item search unit 22, extracts the corresponding combined product items, and outputs the identification information of the combination to the control unit 21 as a second combined product item. In addition, the control unit 21 reads out image data corresponding to the identification information of multiple combinations of the identification information of the similar product item input from the combination search unit 23 and the identification information of the second combined product item from the combination product image database 30 in correspondence with each identification information, and transmits it to the user terminal 200. When the user inputs the identification information of the combined product item in the combination selected by the user, the control unit 21 outputs the received identification information to the product search unit 24.

[0054] The product search unit 24 searches the product item database 26 for a sales product item that is actually on sale and corresponds to the input identification information of the second combination product item, and selects it as a recommended sales product item of a different category to be combined with the sales product item purchased by the user.The product search unit 24 reads image data of this recommended sales product item from the product item image database 27 using the identification information and transmits it to the user terminal 200. In addition, the product search unit 24 may compare the price of the sales product item designated as a recommended product with the price of the sales product item purchased by the user, determine whether it is within a set range (for example, within 0.5 to 2 times the price of the combined product purchased), and transmit the recommended product to the user terminal 200 if it is less than the purchase price, or not transmit the recommended product if it is outside the range.

[0055] Next, the operation of the search support system according to this embodiment will be described with reference to Fig. 7 and Fig. 18. Fig. 18 is a flowchart showing an example of the operation of the search support system in Fig. 7. In the following description, the product input by the user is taken as the product item, the products actually sold in the virtual shop are taken as the sales product items (product items in the product item database 26), and the products extracted from the image media used when selecting a product combination are taken as the combined product items (product items in the combined product image information database 29). When the user terminal 200 accesses the virtual shop managed by the product search support device 100 via the information and communication network I in response to a user operation, the control unit 21 transmits to the user terminal 200 information for acquiring gender, age, and fashion trends, as well as a processing program for displaying combination items (described below) and for selecting images displayed. In the user terminal 200, an image display processing unit and a selection processing unit are provided in an internet browser by the processing program, and the image data transmitted from the product search support device 100 are displayed, and editing and selection processing of the image data are performed.

[0056] Then, when the control unit 21 receives from the user terminal 200 answer data for the gender, age group, and fashion trend entered into the above screen data (for example, data selected from the multiple options displayed on the screen), it selects a corresponding table from the multiple tables in the product item database 26 that corresponds to the gender, age group, and fashion trend, and transmits category information indicating the category of the product item for sale, for example, text information such as shirts, trousers, jackets, shoes, or image information, to the user terminal 200 via the transmission / reception unit 25 and the information communication network I. When text or image information is input, the user terminal 200, in accordance with the processing program, displays categories based on the text or image information (e.g., shirts, trousers, jackets, etc.) on the display unit, and prompts the user to select which item to purchase (e.g., "Please select"). When the user selects any one of the categories, the user terminal 200 transmits category information indicating the category selected by the user, for example, category information indicating shirts, to the product search support device 100 (step F1).

[0057] That is, when category information of a product item that the user wishes to purchase is received, the control unit 21 selects the table shown in FIG. 8 in which shirts are the category (category 1) from the tables of the product item database 26 shown in FIG. 8 to FIG. 10, reads all of the identification information A1 to An of the sales product items in this table, reads out image data of the sales product items with identification numbers A1 to An from the product item image database 27, and transmits the image data of thumbnail images to the user terminal 200 with the identification numbers of the sales product items corresponding to each thumbnail image added. When the image data of the thumbnail images is supplied, the user terminal 200 displays the thumbnail images of the sales merchandise items with the identification numbers A1 to An on the display unit (step F2). Then, when the user selects one of the displayed thumbnail images and inputs usage information, the user terminal 200 transmits identification information of the selected thumbnail image and the usage information to the product search support device 100.

[0058] Upon receiving the identification information and usage information of the selected thumbnail image, the control unit 21 processes the order for the sales product item with this identification number, i.e., the shirt, which is the sales product item selected by the user from the thumbnail image as the item to be purchased, and transmits the input identification number to the similar item search unit 22. Then, when the identification number A1 of a sales product item purchased by a user is input, the similar item search unit 22 extracts product items that match this identification number A1 or have the identification number A1 written on a similar product item and have the same usage information from the tables of the combined product image information database 29 shown in Figures 15 to 17 as a candidate group consisting of one or more product items (step F3). Note that the product image information database 29 also includes products that have the identification information A1 written in the identification information column and A1 written in the similar product item column, and are the same as the products written in the product item database 26. Here, when the similar item search unit 22 receives an identification number A1 of a sales product item from, for example, the combined product image information database 29, it extracts the combined product items with the above identification numbers a1-1 to a1-q, which have the identification number A1 written in the similar product item column in the attribute data, as a candidate group of similar combined product items, and outputs them to the combination search unit 23.

[0059] The combination search unit 23 searches the combination information database 28 shown in FIG. 14 for combined product items of a different category, for example, trousers, which correspond to each of the combined product items with identification numbers in the candidate group input from the similar item search unit 22, extracts corresponding combined product items of trousers (for example, b1-2, b2-3, b1-4, ... corresponding to a1-1, a1-2, a1-3, ... shown in FIG. 14), and outputs the identification information of each of the combinations to the control unit 21 as a second combined product item (step F4). Here, the category recommended as a sales item to be combined with the sales item purchased by the user may be set in advance in correspondence with the purchased item, or the user may be initially allowed to select from a plurality of categories. Here, the combination search unit 23 extracts sales product items to be combined with the sales product item purchased by the user, and then selects from the extracted sales product items to be combined a sales product item that matches the usage information set in the condition setting unit 32, and may set this selected sales product item as the final sales product item to be combined. Then, the control unit 21 reads out image data corresponding to each of the identification information of a plurality of combinations (the number of combined product items corresponding to the candidate group) of the identification information of the combined product item input from the combination search unit 23 and the identification information of the second combined product item from the combined product image database 30 in correspondence with each identification information, adds each identification information to the image data of each combined product item, and transmits it to the user terminal 200.

[0060] When identification information of multiple combinations of identification information of a combined product item (corresponding to a shirt) and identification information of a second combined product item (corresponding to pants) and corresponding image data are input, the user terminal 200 displays the image data of each combination on the display unit (step F5). Also, at the edge of the display screen, a plurality of 3D human images corresponding to different body types created by CG (Computer Graphics) are displayed, and by selecting one of them, the user terminal 200 displays the shirt and pants of the combined product item superimposed on the human image. Also, the user terminal 200 may perform image processing in which the user's own face is captured by an imaging device (not shown) using a web camera mounted on the user's device, and the face area selected by the user from the captured image data is superimposed on the face part of the CG human image.

[0061] Next, the user selects one of the multiple combinations displayed on the display unit of the user terminal 200 using a pointing device or the like (step F6). For example, if the user selects the combination of identification information a1-1 and identification information b1-2, the user terminal 200 transmits the identification information b1-2 of the second combined product item (corresponding to trousers) in the combination selected by the user to the product search support device 100. When the identification information b1-2 of the second combined product item is input via the control unit 21, the product search unit 24 searches the table shown in FIG. 9 corresponding to the trousers category in the product item database 26 (step F7), and extracts the sales product item having the identification information B1 corresponding to this identification information b1-2 (step F8). Then, the control unit 21 searches the product item image database 27 for image data corresponding to the identification information B1 of the combined sales product item extracted by the product search unit 24, and reads it out. Furthermore, the control unit 21 searches and reads image data corresponding to the identification information A1 purchased by the user from the product item image database 27, and transmits this to the user terminal 200 together with image data of the sales product item with the identification information B1.

[0062] When the image data of the sales merchandise items is supplied from the merchandise search support device 100, the user terminal 200 displays the supplied image data of the combination on the display unit (step F9). At this time, similar to step F5 described above, a plurality of 3D human images corresponding to different body types created by CG are displayed at the edge of the display screen, and by selecting one of these 3D human images (for example, an image of a human with a body type similar to the user's own), the user terminal 200 displays the shirt (identification information A1) and pants (identification information B1) of the sales product items superimposed on the selected 3D human image. The user terminal 200 may also capture an image of the user's face using a webcam installed therein, and display the facial area selected by the user superimposed on the facial portion of the human image. Then, when the user selects from the options on the display screen whether or not to purchase the recommended product for sale, which is the identification information B1 (step F10), the user terminal 200 transmits the identification information A1 and identification information B1 to the product search support device 100 along with information indicating whether or not to purchase the recommended product. When the identification information A1 and the identification information B1 are input together with information indicating whether or not to purchase, the control unit 21 performs the order processing in the same manner as in the case of the sales product item of the identification information A1 if the item is to be purchased.

[0063] Furthermore, the product search support device 100 may be provided with a history database 31 indicated by the dashed line in FIG. The history database 31 is made up of a user table having the table structure shown in FIG. 19, and a purchase history table for each user having the table structure shown in FIG. The user table in Figure 19 is configured by associating user identification information that identifies each user, which is assigned to each user who has registered as a member or purchased a sales item, with at least the user's name and user email address corresponding to this user identification number. The purchase history table in Fig. 20 is provided for each user, is identified by the above user identification information, corresponds to the access date and time when the virtual shop was accessed, and stores, as history, identification information of sales product items purchased at the access date and time, identification information of sales product items searched for but not purchased, and identification information of combination product items recommended for the purchased sales product item but not purchased. If there is no purchased sales product item, "-" is stored. The process of writing data to this user table and purchase history table is performed by the control unit 21 as shown below.

[0064] For example, when a user accesses a virtual shop from the user terminal 200 and registers as a member by entering his / her name and email address on a member registration page, the control unit 21 assigns user identification information to the user and additionally registers the user in the user table. The control unit 21 then searches for sales product items in the category selected by the user, and if the sales product item is purchased, stores the purchased sales product item in an area for identification information of the purchased product item, whereas if the sales product item is searched for but not purchased, stores identification information of the searched sales product item in an area for identification information of the searched but not purchased product item. In addition, when a user purchases a sales product item as a recommended product to be combined with the purchased sales product item, the control unit 21 stores the purchased sales product item in an area of ​​the purchase history table for identification information of the purchased product item, and on the other hand, when the user does not purchase the sales product item as the recommended product, the control unit 21 stores the purchased sales product item in an area for identification information of the recommended but not purchased product item.

[0065] Then, when the user accesses the virtual shop again and searches for a product item for sale, the control unit 21 selects the user's purchase history table from the history database 31 based on the user identification information entered by the user, and searches the selected purchase history table based on the identification information of the product item for sale. At this time, if the control unit 21 detects the identification information of the sales product item that the user is searching for in the area of ​​the identification information of the purchased product item, it adds the date and time of purchase of the same sales product item, transmits it to the user terminal 200, and displays it on the display screen to notify the user of the combined commercial items that have already been purchased.

[0066] Furthermore, when the control unit 21 detects the identification information of the sales product item being searched for in the area of ​​the identification information of a product item that was searched for but not purchased, it adds the date and time of the search for the same sales product item and transmits it to the user terminal 200, displays it on the display screen, and notifies the user of information that stimulates the user's desire to purchase, such as "This is a product item you searched for before, and isn't it your favorite type?" Furthermore, when the control unit 21 detects the identification information of the sales item being searched for in the area of ​​the identification information of a product item that was recommended but not purchased, it adds the date and time when the same sales item was recommended and image data of the sales item purchased at that time, transmits it to the user terminal 200, displays it on the display screen, and notifies the user of information that stimulates the user's desire to purchase, such as "This is a good product item to combine with a product item you purchased previously."

[0067] Next, the feature data will be described. As shown in Fig. 8 to Fig. 10 and Fig. 15 to Fig. 17, the feature data is, for example, a two-dimensional Fourier transform of the pattern of the fabric of the clothing for each color space of R (red), G (green), and B (blue). The control unit 21 generates the element data RD, GD, and BD of this feature data by two-dimensional Fourier transform. At this time, if the sales product item or combined product item is a shirt or jacket, the width or shoulder width of the clothing is used as the reference value for the length when performing the two-dimensional Fourier transform, so that data consistency is achieved in the processing between each sales product item and combined product item. In the case of trousers, the width of the waist part is used as the reference value for length when performing a two-dimensional Fourier transform, so as to ensure data consistency in processing between each sales product item and combined product items.

[0068] That is, in order to confirm similarity using feature data and accurately determine the size of the pattern, etc., it is necessary to standardize the overall dimensions of each part using the dimensions of any position on the accessory as a reference value so that the results of the two-dimensional Fourier transform of combined products in the same category can be consistent.When capturing image data for the two-dimensional Fourier transform, the shirt, jacket, trousers, etc. are placed on a flat floor and, after flattening, are captured by a digital camera, etc. Moreover, the shape as element data of the feature data is, for example, the ratio of the sleeve length to the shoulder width in the case of a shirt or jacket, or the ratio of the crotch width to the hem width in the case of trousers. Next, for the texture, the largest part of the fabric is enlarged and a two-dimensional Fourier transform is performed on the shape of the fabric. At this time, the enlargement rate is kept constant and the consistency of the two-dimensional Fourier transform data obtained from the image data of different combination items is ensured.

[0069] As described above, a salesperson at the virtual store takes image data of clothing, and the control unit 21 performs Fourier transforms, etc., on this captured image data as described above to collect feature data from the image data of the product items being sold, and also collect attribute data from image data of combined product items collected from image media such as fashion catalogs or the Internet. Then, the control unit 21 performs clustering of the combined product items collected from the image media, with the feature data of the sales product items actually being sold as the center of gravity data of the clusters, for the combined product items having similarity. Here, the control unit 21, for example, calculates the distance between the comparison feature vector consisting of the feature data of the image data of each combined product item collected from the image media input by the store clerk and the reference feature vector consisting of the feature data of the image data of each sales product item actually being sold, performs processing to classify the combined product items having feature vectors closer to each sales product item than the reference feature vectors of the other combined product items as clusters having similarity with the sales product items that are considered to be close, and generates the correspondence relationship between the combined product items and the sales product items in the column of similar product items in the tables of Figures 15, 16, and 17.

[0070] Similarly, a correspondence relationship is generated between the sold product items and the combined product items in the similar product items column in the tables of Figures 8, 9, and 10. That is, the identification information described in the similar product items column in the tables of Figures 8, 9, and 10 is the identification information of a combined product item that has a shorter distance in feature vector, i.e., similarity, to the sold product item of the corresponding identification information compared to other sold product items. Here, the control unit 21 may store the similar product items in the columns of Figures 8, 9, and 10 in order of shortest distance of feature vectors, i.e., in order of highest similarity. This allows the similar item search unit 22 to easily extract the most similar combined product item or the hth (h is preset) most similar combined product item from the combined product items collected from the image medium from Figures 15 to 17 when selecting the combined product item most similar to the sales product item.

[0071] Furthermore, as described above, clustering may be performed in advance by comparing the reference feature vector of each product item for sale with feature data of combined product items collected from image media having comparative feature vectors that are close to each other, and instead of providing a column for recording the identification information of similar product items, the similarity may be calculated each time a similar combined product item is searched for. For example, in the search for combined product items obtained from image media similar to the sales product item in step F3, the similar item search unit 22 may be configured to calculate the similarity (the shorter the distance between the feature vectors, the higher the similarity) based on the distance between the above-mentioned reference feature vector and the comparison feature vector, and extract combined product items collected from image media that have similarity to the sales product item.

[0072] At this time, similarly, in searching for sales product items similar to the combined product items acquired from the image media in step F7, the product search unit 24 may be configured to calculate the similarity based on the distance between the above-mentioned reference feature vector and the comparison feature vector, and extract sales product items that have similarity to the combined product items collected from the image media. Furthermore, the similar item search unit 22 and the product search unit 24 may calculate the distance between the sold product item and the combined product items collected from the image media, and extract the one with the highest similarity or the one with the highest similarity up to the h-th similarity. As described above, according to this embodiment, when a user searches for an object, the search can be supported by providing information or the like. For example, as described above, when purchasing a combination product, products that match the product the user is going to purchase or has already purchased are extracted from a group of products on sale, and the extracted products are recommended to the user, so that the user can easily select a combination product without having to worry about whether a product is suitable or not when selecting a suitable product from many products available in a virtual store on the Internet.

[0073] In the above-described embodiment, the use information is added as one of the attribute information of each product item in the combined product image information database 29, but the combined product image information database 29 may be configured for each category of use information. For example, product tables for spring, summer, autumn, and winter may be configured and used. In this case, the similar item search unit 22 selects the product image information database 29 corresponding to the registration information added to the search request from the user terminal 200, and searches for a product item similar to the product item requested to be searched from the selected product image information database 29, thereby making it possible to shorten the search time.

[0074] <Third embodiment> In the second embodiment, the product search support device 100 transmits sales product items as thumbnail images to the user terminal 200, and allows the user to select an item on the screen of the user terminal 200. In the third embodiment, after a user accesses the virtual shop, image data of the clothing he or she wishes to purchase, read from an image medium using a scanner or the like, or downloaded from the Internet, may be input as image data of a product item to the user terminal 200, which then transmits this image data to the product search support device 100, which then extracts sales product items similar to the combined product items of this image data. The operations after the sales product items are extracted are the same as those after the user selects a sales product item from a thumbnail image in the second embodiment.

[0075] The configuration of the product search support device 100 according to the third embodiment is similar to that of the second embodiment. Only the operations different from the second embodiment will be described below. When a user accesses the virtual shop via the user terminal 200, the product search support device 100 is started, and the control unit 21 transmits to the user terminal 200 image information of an input screen (where the user searches by clicking on options with a mouse or the like) that asks the user whether to select a sales product item from a thumbnail image, or whether to extract similar sales product items from the product item database 26 based on image data of clothing entered by the user. Then, when the user decides to select an item to purchase from the thumbnail images and a response signal indicating the selection is input from the user terminal 200 using the thumbnail images, the control unit 21, like the second embodiment, adds identification information of the sales item to each image data and transmits thumbnail images of the multiple sales item items to the user terminal 200. After the user selects one of the thumbnail images, the process is the same as in the second embodiment.

[0076] On the other hand, when the user selects to extract similar sales product items from the product item database 26 based on image data input by the user, the user terminal 200 transmits a response signal indicating that the selection is based on the image data input by the user to the product search support device 100. In response to this, the control unit 21 transmits information on the input screen for inputting image data to the user terminal 200. The user loads image data of the clothing he or she wishes to purchase (paper image media such as fashion magazines or fashion catalogs) into user terminal 200 using a scanner, or takes a picture of the clothing with a digital camera and loads the captured image data, or image data obtained via the Internet, into user terminal 200. When the image data is read, the image data is displayed on the display unit in the image data display area of ​​the input screen of the user terminal 200 as shown in Fig. 21, and the user selects the clothing part of the image data that he / she wishes to purchase, for example, the area of ​​the shirt part that he / she wishes to extract, with a dashed line H1, and selects the background image part of the entire shirt part that he / she wishes to extract by surrounding the outer periphery of the shirt part that he / she wishes to extract (the part near the outer periphery of the shirt) with a dashed line H2. Then, the user selects a category name from a list of categories (for example, shirts) in a combo box on the display screen of the display unit of the user terminal 200, and when the user clicks the send button on the display screen, the user terminal 200 detects that a process of sending the image data has been requested, and sends the image data of the clothing on which the dashed lines H1 and H2 are drawn, the image data of the clothing on which the dashed lines are not drawn, and category information indicating the category name of the selected clothing to the product search support device 100.

[0077] Then, when the control unit 21 receives image data of clothing with dashed lines H1 and H2 drawn and image data of clothing without dashed lines drawn, the control unit 21 performs a process of segmenting the image data of the clothing into the shirt portion and other portions using the Graph-Cut method, and extracts the shirt portion. That is, the control unit 21 performs segmentation by determining the boundary between an area having pixel values ​​similar to the pixels along dashed line H1 drawn on the shirt portion and an area having pixel values ​​similar to the pixels along dashed line H2 drawn on an area other than the shirt portion as the position where the error of the error function due to gradient is minimum. Here, when the control unit 21 extracts the shirt portion, it performs a two-dimensional Fourier transform as described above, and also extracts shape data as an element in the feature data, and outputs it as detection target feature data.

[0078] Then, the control unit 21 selects a table corresponding to the extracted clothing category from the product item database 26, for example, a table of shirts in this embodiment. After selecting the table of shirts, the control unit 21 calculates the distance between a reference feature vector consisting of feature data of each sales item in the table and a target feature vector consisting of the detection target feature data, and extracts from the table of shirts the top h sales item(s) that have feature data similar to the detection target feature data that is equal to or less than a preset distance and that are determined to be similar, for example, the top 5. Here, since it is difficult to extract texture from image data sent by a user in the feature data, when generating a feature vector from the feature data, the control unit 21 excludes texture from the elements of the vector and calculates the distance between the two feature vectors to be compared. Next, the control unit 21 reads image data of each of the top five sales product items from the product item image database 27 using the identification information of the top five products, and transmits the thumbnail images corresponding to the respective identification information to the user terminal 200. The subsequent processing is the same as the processing after transmitting the thumbnail images in step F2 in the flowchart of Fig. 18 in the second embodiment.

[0079] <Fourth embodiment> In the third embodiment, a sales item that is identical or similar to the image data of the product item entered by the user is searched for in a product item database 26 that stores sales product items that are on sale, and the searched sales item is presented to the user, while also recommending sales items of other categories to be combined with the searched sales item. In the fourth embodiment, an operation is performed to search for sales product items to be combined with clothing already owned by the user.

[0080] A user takes an image of his / her own clothing, for example a shirt, with a digital camera, and causes the user terminal 200 to read the image data of the shirt from the digital camera. When the user accesses the virtual shop via the user terminal 200, the product search support device 100 is started, and the control unit 21 transmits to the user terminal 200 image information of an input screen (where the user searches by clicking options with a mouse or the like) that asks whether the user should select a sales product item from a thumbnail image, or whether similar sales product items are extracted from the product item database 26 based on image data of clothing entered by the user, or whether the user should recommend a sales product item to be combined with the clothing in the image data entered by the user.

[0081] Then, when the user decides to select an item to purchase from the thumbnail images and a response signal indicating the selection is input from the user terminal 200 using the thumbnail images, the control unit 21, like the second embodiment, adds identification information of the sales item to each image data and transmits thumbnail images of the multiple sales item items to the user terminal 200. After the user selects one of the thumbnail images, the process is the same as in the second embodiment. On the other hand, when the user selects to extract sales product items from the product item database 26 based on image data input by the user, the user terminal 200 transmits a response signal indicating that the selection is based on image data input by the user to the product search support device 100. The subsequent processing is the same as that in the third embodiment.

[0082] In addition, when the user selects to recommend sales product items of other categories to be combined with the clothing in the image data input by the user, the user terminal 200 transmits a response signal to the product search support device 100 indicating that the user has selected to recommend sales product items to be combined with the clothing in the image data input by the user. Hereinafter, a process of recommending sales product items to be combined with clothing in image data input by a user according to the fourth embodiment will be described. The configuration of the product search support device 100 according to the fourth embodiment is the same as that of the second embodiment. Hereinafter, only operations different from the first and third embodiments will be described.

[0083] As described above, when the user selects to recommend a sales product item to be combined with the clothing of the image data input by the user, the user terminal 200 transmits a response signal indicating that the sales product item to be combined with the clothing of the image data input by the user is recommended to the product search support device 100. In response to this, the control unit 21 transmits information of the input screen for inputting the image data and the use information to the user terminal 200. The user takes image data of an outfit (for example, a shirt) that he or she owns and wants to have a recommended sales item combined with it, using an imaging device such as a digital camera, and loads the image data of the outfit from the imaging device into the user terminal 200.

[0084] When the image data loaded by the user terminal 200 is displayed on the display unit, the user selects from a combo box the clothing category of the image data displayed in the image data display area and inputs it to a category input field located near the image data display area. The user also selects from a combo box and inputs usage information located near the image data display area on the display unit of the user terminal 200. Then, when the user selects the send button on the input screen using a pointing device such as a mouse, the user terminal 200 detects that a process to send and process the image data has been requested, and sends the image data displayed in the image data display area to the product search support device 100 together with category information indicating the above category and use information. When image data is input, the control unit 21 performs a two-dimensional Fourier transform on the clothing of the image data, for example, image data of a shirt, extracts shape data as an element in the feature data, and outputs this together with category information as detection target feature data to the similar item search unit 22. The control unit 21 also outputs the use information supplied together with the image data and category information to the condition setting unit 32. As a result, the condition setting unit 32 internally sets the use information used when searching for similar product items, and outputs it to the similar item search unit 22.

[0085] Then, the similar item search unit 22 selects a table corresponding to the extracted clothing category, for example, a table of shirts, from the combination product image information database 29 . After selecting the shirt table, the similar item search unit 22 calculates the distance between the reference feature vector consisting of the feature data of each combined product item in the table and the target feature vector consisting of the detection target feature data, and extracts from the shirt table the top h combined product items, for example the top 5 combined product items, having feature data similar to the detection target feature data that is equal to or less than a preset distance, that match the usage information set in the condition setting unit 32 and are determined to be similar in order of shortest distance. Here, since it is difficult to extract texture from image data sent by a user in the feature data, when generating the feature vector from the feature data, the control unit 21 removes texture from the elements of the vector to generate the reference feature vector and the detection target feature vector.

[0086] Then, the similar item search unit 22 outputs the identification information of the detected combined product items having the top five similarities, for example, five shirts, to the combination search unit 23. When the identification information of the combined product items is input, the combination search unit 23 reads out the identification information of combined items of other categories, for example, trousers, stored in the combination information database 28 in correspondence with each of the five identification information of the input combined product items. Next, the combination search unit 23 transmits the combinations of identification information for each of the five combinations of shirts and trousers to the control unit 21. The subsequent processing is the same as the processing from step F5 onwards in the flowchart of FIG.

[0087] <Fifth embodiment> The combined product image information database 29 may store attribute data of image data of old clothing collected from old fashion magazines or fashion catalogs from the past, for example, 10 or 20 years ago, in association with identification information. Moreover, the combined product image database 30 stores image data of old clothing in association with the identification signals. Then, a table of combinations of combined product items between different categories shown in FIG. 14 is created in the combination information database 28 in accordance with combinations of combined product items found in past fashion magazines or fashion catalogs. As described above, by configuring the combined product image information database, the combined product image database 30 and the combination information database 28, when a new sales product item similar to a past design is sold, it becomes possible to easily extract sales product items to be combined with the new combined product item similar to the past design from the sales product items currently being sold, by utilizing the combinations in the past designs. In this embodiment, clothing has been described as an example of a product item, but the present invention can be easily applied to combination products that combine products of multiple different types (categories), such as combination furniture, combination home appliances, and other products in general.

[0088] In each of the above embodiments, a program for implementing the functions of each part of the search server 1, 1a, and the product search support device 100 may be recorded in a computer-readable recording medium, and the program recorded in the recording medium may be read into a computer system and executed to perform product search support processing. The term "computer system" as used herein includes hardware such as an OS and peripheral devices. The term "computer system" also includes a WWW system equipped with a homepage providing environment (or display environment). The term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, and storage devices such as hard disks built into a computer system. The term "computer-readable recording medium" also refers to a storage device that holds a program for a certain period of time, such as a volatile memory (RAM) inside a computer system that becomes a server or client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line.

[0089] The above program may also be transmitted from a computer system in which the program is stored in a storage device or the like to another computer system via a transmission medium, or by transmission waves in the transmission medium. Here, the "transmission medium" that transmits the program refers to a medium that has the function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. The above program may also be for realizing part of the above-mentioned functions. Furthermore, it may be a so-called difference file (difference program) that can realize the above-mentioned functions in combination with a program already recorded in the computer system. As described above, according to the aspects of the present invention, a user can easily search for an object that he or she wishes to search for or an object similar to that object. [Industrial Applicability]

[0090] The present invention is suitable for use in a search support system or similar technology that supports a user in selecting or determining an object to search for when searching for the object via the Internet, and can perform a search for the object that the user wishes to search for using image data of the object. [Explanation of symbols]

[0091] 1...Search server 2. User terminal 11...Extraction part 12...Type determination section 13…Search section 14...Combination extraction section 15...Preference extraction part 16…Database 17...Transmitter / receiver 18, 32...Condition setting section 21...Control unit 22…Similar Item Search Section 23…Combination search section 24…Product Search Section 25…Transmitter / receiver 26…Product Item Database 27…Product item image database 28…Combination information database 29…Combined product image information database 30…Combined product image database 31...Historical database 100...Product search support device 200...User terminal I…Information and communication network

Claims

1. A database in which object image data relating to an object searched for by a user is stored; an extraction unit for extracting a search portion of input image data; a search unit that compares image feature information indicating features of an image of the image data of the search portion with image feature information indicating features of an image of the object image data in the database, and retrieves from the database the object image data that matches or is highly similar to the image data of the search portion; a condition setting unit that extracts search conditions including at least one of date and time information, location information, and situation information based on the input image data, and sets the extracted search conditions; The search unit acquires the object image data based on the search conditions set by the condition setting unit.

2. A type determination unit that determines a type of the object corresponding to the search portion, 2. The search support system according to claim 1, wherein the search unit acquires the object image data based on the type determined by the type determination unit.

3. The database stores object information corresponding to the object image data, 3. The search support system according to claim 1, wherein the object information corresponding to the object image data acquired by the search unit is output.

4. 4. The search support system according to claim 3, wherein the object information includes at least one of a name, a sales store, and a price.

5. The search support system according to any one of claims 1 to 4, characterized in that, together with the object image data acquired by the search unit, the object image data of an object other than the object to be searched is output.

6. 6. The search support system according to claim 5, wherein the type of the different object is the same as the type of the object to be searched.

7. 6. The search support system according to claim 5, wherein the type of the different object is different from the type of the object to be searched.

8. a combination extraction unit that extracts, from the database, image data of another type of object that can be combined with the object corresponding to the object image data acquired by the search unit; 8. The search support system according to claim 1, further comprising: a step of outputting the object image data extracted by the combination extraction unit.

9. 9. The search support system according to claim 8, wherein the combination extraction unit extracts the object image data based on the search conditions set by the condition setting unit.

10. The combination extraction unit includes: an image of the object image data acquired by the search unit; generating composite image data by combining the extracted object image data with the image data; 10. The search support system according to claim 8, further comprising: a step of outputting the composite image data generated by the combination extraction unit.

11. a combination extraction unit that extracts, from the database, the object image data of other types of objects that are likely to be purchased in combination with the object corresponding to the object image data acquired by the search unit; 11. The search support system according to claim 1, further comprising: a step of outputting the object image data extracted by the combination extraction unit.

12. A user terminal which is a terminal of a user; A server device, The extraction unit is provided in the user terminal, the database and the search unit are provided in the server device, 12. The search support system according to claim 1, wherein the user terminal has a transmitting / receiving unit that transmits image data of the search portion extracted by the extracting unit to the server device.

Citation Information

Patent Citations

  • On-line shopping system and on-line shopping method

    JP2002150138A

  • Retrieval guide system, terminal and server

    JP2003122757A

  • Fashion merchandise sales system and fashion merchandise sales program

    JP2004246585A

  • Price information retrieval device, price information retrieval system and price information retrieval method

    JP2008052672A