Item search system, item search method, and item search program

Abstract

An item search system, an item search method, and an item search program are provided for allowing a user to search for a desired item. [Solution] The support device 20 includes a control unit 21 connected to the user device 10. The control unit 21 executes an item acquisition process to acquire a search target item from the user device 10, calculates image features of the search target item, identifies a specified number of similar items from an item information storage unit 22 based on the image features, classifies the specified number of similar items into multiple groups, identifies a representative item from the similar items belonging to each group, and outputs the representative item for each group to the user device 10, and repeats the item search process using the representative item selected in the user device 10 as the search target item.

Description

[Technical Field]

[0001] The present disclosure relates to an item search system, an item search method, and an item search program for searching for items such as images according to user preferences. [Background technology]

[0002] Information from a user may be acquired to search for items that match the user's preferences. For example, a technology has been studied that, for an input image as a query, retrieves images of the same subject from pre-registered reference images (see, for example, Patent Document 1). In the technology described in this patent document, an index is created based on features extracted from the reference images. Features are then extracted from the input query image. By referencing the index and adding a candidate image score to reference images with features that belong to the same cluster, images with a high candidate image score are selected as candidate images. For each candidate image, a detailed score is calculated by directly referencing the features of the reference images and performing more detailed recognition processing based on the similarity with the features of the query image. The reference image with a high total score, calculated by weighting the normalized values ​​of the candidate image score and the detailed score, is output as a similar image.

[0003] An item search system for searching for a user's desired item has also been studied (for example, Patent Document 2). The support server described in this patent document includes a control unit connected to a user device. The control unit outputs to the user device a first item consisting of a first component value and a plurality of item candidates consisting of component values ​​different from the first component value in a plurality of principal components that make up the item. A second item selected by the user device is then identified from the item candidates. Next, for each principal component, a positional relationship between the first component value and the second component value of the second item is calculated. Furthermore, for each principal component, a component value distribution is calculated according to the positional relationship. A plurality of item candidates are then newly generated for the second item based on the component value distribution and output to the user device. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2017-016501 [Patent Document 2] Japanese Patent Publication No. 2023-002325 Summary of the Invention [Problem to be solved by the invention]

[0005] However, images contain a variety of elements, such as shape, pattern, color, and atmosphere. When a user selects an image, the elements that determine the image are also diverse. Furthermore, the elements desired by the user may change during the process of selecting candidate items. It may be difficult to present appropriate item candidates to such a user. [Means for solving the problem]

[0006] An item search system that solves the above problem includes a control unit connected to a user device and an item information storage unit storing a plurality of items. The control unit executes an item acquisition process to acquire a search target item from the user device, calculates image features of the search target item, identifies a specified number of similar items from the item information storage unit based on the image features, classifies the specified number of similar items into a plurality of groups, identifies a representative item among the similar items belonging to each group, and outputs the representative item for each group to the user device. The control unit then repeats the item search process using the representative item selected on the user device as the search target item. [Effects of the Invention]

[0007] According to the present disclosure, a user can efficiently and accurately search for a desired item. [Brief explanation of the drawings]

[0008] [Figure 1] FIG. 1 is an explanatory diagram of an item search system according to a first embodiment. [Figure 2] FIG. 2 is an explanatory diagram of a hardware configuration of the first embodiment. [Figure 3] FIG. 4 is an explanatory diagram of a processing procedure of the selection support processing according to the first embodiment. [Figure 4] FIG. 2 is an explanatory diagram of a representative item selection method according to the first embodiment. [Figure 5] FIG. 10 is an explanatory diagram of a processing procedure according to a second embodiment. [Figure 6] FIG. 10 is an explanatory diagram of a processing procedure of another example of selection support processing. [Figure 7] FIG. 10 is an explanatory diagram of another example of a process for narrowing down a search range. [Figure 8] FIG. 10 is an explanatory diagram of another example of a selection history. [Figure 9] FIG. 10 is an explanatory diagram of an extraction range according to a selection history in another example. [Figure 10] FIG. 10 is an explanatory diagram of another example of a representative item selection method. [Figure 11] 10A and 10B are explanatory diagrams of another example of processing procedures, in which FIG. 10A is an explanatory diagram of the attribute feature axis generation processing, and FIG. 10B is an explanatory diagram of the search range narrowing processing. [Figure 12] 10A and 10B are explanatory diagrams of another example of a search range, in which (a) is the same color, and (b) is an explanatory diagram of color variations. DETAILED DESCRIPTION OF THE INVENTION

[0009] (First embodiment) An embodiment of an item search system, an item search method, and an item search program will be described with reference to FIGS. 1 to 4. In this embodiment, it is assumed that a user searches for a favorite item according to their taste. Here, prepared two-dimensional still images are used as items. In this embodiment, the two-dimensional still images include photographed images of clothes that can be provided to the user. As shown in FIG. 1, the item search system of this embodiment uses a user device 10 and a support device 20 connected via a network.

[0010] (Example of hardware configuration) FIG. 2 shows an example of the hardware configuration of an information processing device H10 that functions as the user device 10, the support device 20, and the like.

[0011] The information processing device H10 includes a communication device H11, an input device H12, a display device H13, a storage device H14, and a processor H15. Note that this hardware configuration is an example, and the information processing device H10 may include other hardware.

[0012] The communication device H11 is an interface that establishes a communication path with another device and executes data transmission and reception, and is, for example, a network interface or a wireless interface.

[0013] The input device H12 is a device that accepts input from a user, etc., and is, for example, a mouse, a keyboard, etc. The display device H13 is a display, a touch panel, etc. that displays various information.

[0014] The storage device H14 is a storage device that stores data and various programs for executing various functions of the user device 10 and the support device 20. Examples of the storage device H14 include a ROM, a RAM, and a hard disk.

[0015] The processor H15 controls each process (for example, the process in the control unit 21, which will be described later) in the user device 10 and the support device 20 using programs and data stored in the storage device H14. Examples of the processor H15 include a CPU and an MPU. The processor H15 loads a program stored in a ROM or the like into a RAM and executes various processes corresponding to the various processes. For example, when an application program in the user device 10 or the support device 20 is started, the processor H15 runs a process that executes each process, which will be described later.

[0016] The processor H15 is not limited to a processor that performs all of its processing using software. For example, the processor H15 may include a dedicated hardware circuit (e.g., an application-specific integrated circuit (ASIC)) that performs hardware processing for at least some of the processing it performs. That is, the processor H15 may be configured with the following:

[0017] (1) One or more processors that operate according to a computer program (2) One or more dedicated hardware circuits that perform at least some of the processes; or (3) A circuit including a combination thereof The processor includes a CPU and memory, such as RAM and ROM, that stores program code or instructions configured to cause the CPU to perform processes. Memory, or computer-readable media, includes any available media that can be accessed by a general-purpose or special-purpose computer.

[0018] (Functions of each information processing device) The functions of the user device 10 and the assistance device 20 will be described with reference to FIG. The user device 10 is a computer terminal used by a user of this system.

[0019] The support device 20 is a computer system for supporting a user in searching for a desired item. The support device 20 includes a control unit 21, an item information storage unit 22, and a history information storage unit .

[0020] The control unit 21 performs the processes described below (processes including a management stage, an extraction stage, a candidate output stage, etc.) By executing an item search program for this purpose, the control unit 21 functions as a management unit 211, an extraction unit 212, a candidate output unit 213, etc.

[0021] The management unit 211 executes a process of acquiring and managing information about items. The extraction unit 212 executes a process of identifying similar items based on the items identified in the user device 10. The candidate output unit 213 executes a process of outputting the item candidates identified from among the similar items to the user device 10.

[0022] The item information storage unit 22 stores item management information related to items. This item management information is recorded when an item that can be provided to the user device 10 is registered. This item management information includes data related to an item identifier, an item image, an item attribute, and image features. For example, an image of a piece of clothing is used as an item.

[0023] The item identifier is an identifier for identifying each item. An item image is a photograph of an item (clothing). Item attributes are elements that make up this item (material, color, pattern, shape, size, style, brand, price, function, etc.).

[0024] Image features are the features of this item image. For example, a multidimensional vector calculated from the item image by deep learning using a convolutional neural network (CNN) can be used as the image feature. Alternatively, principal component analysis of the pixels of the item image can be performed, and a multidimensional vector consisting of the principal components of each component can be used as the image feature.

[0025] History management information about items selected by the user is recorded in the history information storage unit 23. This history management information is recorded when the user selects an item. This history information includes data about the initial item and item identifier for each selection.

[0026] The number of selections is the number of times the selection process is repeated (n times). The initial item is an image acquired from the user device 10 (the item to be searched). The item identifier is the identifier of the item selected by the user at the time of selection.

[0027] (Selection support processing) Next, the selection support process will be described with reference to FIG. First, an item acquisition process is performed. Here, the control unit 21 of the support device 20 executes a sample image acquisition process (step S11). Specifically, the management unit 211 of the control unit 21 acquires a sample image from the user device 10 as a preceding item. Then, the management unit 211 records the sample image in the history information storage unit 23 as an initial item (item to be searched).

[0028] Next, an item search process is performed. Here, the control unit 21 of the support device 20 executes a process of extracting a designated number of similar images (step S12). Specifically, the extraction unit 212 of the control unit 21 calculates image feature amounts of sample images. Then, the extraction unit 212 extracts the designated number of pieces of item management information from the item information storage unit 22 using the feature amounts of the sample images. In this case, the extraction unit 212 extracts item management information whose image feature amounts are included in a similarity range within a predetermined distance from the image feature amounts of the sample images.

[0029] Next, the control unit 21 of the support device 20 executes a similar image clustering process (step S13). Specifically, the extraction unit 212 of the control unit 21 performs clustering by cluster analysis process using the image feature quantities of the specified number of extracted item management information to classify the information into a plurality of groups (clusters).

[0030] Next, the control unit 21 of the support device 20 executes a process of identifying a candidate item that represents each cluster (step S14). Specifically, the extraction unit 212 of the control unit 21 identifies a representative item in each cluster according to the positional relationship of similar items. In this case, the method of identifying the representative item is changed according to the positional relationship between the cluster and the preceding item.

[0031] 4, for the nearest cluster C12 to the preceding item, a similar item 501 that is the closest to the preceding item among the similar items belonging to cluster C12 is identified as the representative item (candidate item). For the other clusters C11, C13, and C14, the center position of each cluster C11, C13, and C14 is identified, and a similar item 502 having a feature value close to this center position is identified as the representative item (candidate item). In this way, the extraction unit 212 identifies candidate items for the number of clusters.

[0032] Next, the control unit 21 of the support device 20 executes a candidate item output process (step S15). Specifically, the candidate output unit 213 of the control unit 21 outputs an item selection screen to the display device H13 of the user device 10. This item selection screen includes item images of the identified candidate items, selection buttons for sample images (previous item images), and a selection button for instructing termination.

[0033] Next, the control unit 21 of the support device 20 executes an item selection process (step S16). Specifically, if the user finds a desired item candidate on the item selection screen, the user presses the selection button for the desired item candidate. Furthermore, to end the process, the user presses the end button. If the desired item is not found, the user presses the selection button for the preceding item (e.g., a sample image). In this case, the candidate output unit 213 of the control unit 21 identifies the pressed selection button. When an item candidate is selected, the candidate output unit 213 acquires the item identifier of this item candidate from the user device 10. In this case, the candidate output unit 213 adds "1" to the number of selections in the history management information recorded in the history information storage unit 23.

[0034] Next, the control unit 21 of the support device 20 executes a determination process as to whether or not to end (step S17). Specifically, the candidate output unit 213 of the control unit 21 determines that the process is ended when detecting that the end button is pressed on the item selection screen.

[0035] If it is determined that the selection support process should be ended (YES in step S17), the control unit 21 of the support device 20 ends the selection support process. If it is determined that the process is not finished ("NO" in step S17), the control unit 21 of the support device 20 executes a determination process as to whether the selected item is a new item (step S18). Specifically, the candidate output unit 213 of the control unit 21 determines whether the selected item is a previous item or a new item candidate.

[0036] If it is determined that a new item candidate has been selected ("YES" in step S18), the control unit 21 of the support device 20 executes a process of decreasing the designated number (step S19). Specifically, the extraction unit 212 of the control unit 21 decreases the designated number according to the number of repetitions of the item search process. In this case, for example, the decreased designated number is calculated according to the number of repetitions.

[0037] On the other hand, if it is determined that the preceding item has been selected ("NO" in step S18), the control unit 21 of the support device 20 executes a process of increasing the designated number (step S20). Specifically, the extraction unit 212 of the control unit 21 increases the designated number. In this case, for example, the increased designated number is calculated according to the number of times the item search process is repeated.

[0038] Then, the control unit 21 of the support device 20 repeats the item search process from the extraction process of the designated number of similar images (step S12) onward, extracting images similar to the selected item candidate or preceding item.

[0039] (Operation of the first embodiment) Since the representative items of each cluster into which similar items are classified are used as candidate items, the feature quantities of the candidate items are dispersed.

[0040] (Effects of the first embodiment) (1-1) In this embodiment, the control unit 21 of the support device 20 executes a process of acquiring a sample image (step S11) and a process of extracting a designated number of similar images (step S12). This makes it possible to extract item candidates that resemble the sample image. (1-2) In this embodiment, the control unit 21 of the support device 20 executes a similar image clustering process (step S13) and a process of identifying candidate items that represent each cluster (step S14). Because sample images contain a variety of elements, such as color, pattern, and shape, it may be difficult to determine what elements the user desires in the sample image. Even in this case, it is possible to extract dispersed item candidates in clusters classified by feature amounts. Furthermore, by identifying item candidates that are close to the sample image in the nearest cluster, it is possible to extract item candidates that are similar to the sample image overall.

[0041] (1-3) In this embodiment, the control unit 21 of the support device 20 executes a candidate item output process (step S15) and an item selection process (step S16). This allows the user to identify a candidate item desired from among existing items.

[0042] (1-4) In this embodiment, when it is determined that a new item candidate has been selected ("YES" in step S18), the control unit 21 of the support device 20 executes a process of decreasing the designated number (step S19). This allows item candidates to be identified within the narrowed similarity range.

[0043] (1-5) In this embodiment, when it is determined that the preceding item has been selected ("NO" in step S18), the control unit 21 of the support device 20 executes a process of increasing the designated number (step S20). As a result, when the user cannot find the item candidate desired, it is possible to identify item candidates within the expanded similarity range.

[0044] (Second embodiment) In the first embodiment, items are classified by clustering. The method of classifying items is not limited to clustering. In the second embodiment, a case will be described in which items are classified using principal component analysis instead of clustering.

[0045] The selection support process in this case will be described with reference to FIG. First, the control unit 21 of the support device 20 executes the process of acquiring a sample image (step S21) and the process of extracting a designated number of similar images (step S22), similar to steps S11 and S12.

[0046] Next, the control unit 21 of the support device 20 executes a principal component analysis process on the similar image (step S23). Specifically, the extraction unit 212 of the control unit 21 calculates image feature amount components by performing a principal component analysis on the similar image.

[0047] Next, the control unit 21 of the support device 20 executes a process of identifying feature components with large variations (step S24). Specifically, the extraction unit 212 of the control unit 21 identifies a plurality of feature component axes in the order of the largest variations of the image feature components in the principal component analysis.

[0048] Next, the control unit 21 of the support device 20 executes a process of identifying candidate items using the feature components (step S25). Specifically, the management unit 211 of the control unit 21 identifies candidate items on multiple feature component axes with large variations. Here, among similar images grouped on these component axes, similar items with maximum and minimum values ​​are identified as candidate items. Next, the control unit 21 of the support device 20 executes a process of outputting candidate items (step S15) to a process of increasing the designated number (step S20).

[0049] (Effects of the second embodiment) According to this embodiment, in addition to the effects (1-1) and (1-3) to (1-5), the following effects can be obtained.

[0050] (2-1) In this embodiment, the control unit 21 of the support device 20 executes a principal component analysis process of similar images (step S23), a process of identifying feature components with large variations (step S24), and a process of identifying candidate items using the feature components (step S25). This makes it possible to identify candidate items consisting of representative items according to the variations in similar images.

[0051] This embodiment can be modified as follows: This embodiment and the following modifications can be combined and implemented within the scope of technical compatibility. In the above embodiments, it is assumed that a two-dimensional still image is searched for as an item. The search target is not limited to a two-dimensional still image. It can be applied to anything that can quantify the elements that make up the item in multiple dimensions. For example, it can be applied to three-dimensional images, videos, audio, and text.

[0052] In each of the above embodiments, the control unit 21 of the support device 20 executes a candidate item output process (step S15). Here, item candidates that are close in distance in terms of feature quantities of each item candidate are displayed close together. This makes it easier for the user to compare the item candidates.

[0053] In each of the above embodiments, item candidates are identified in groups classified by clustering or principal component analysis. In addition to this, the user may be allowed to narrow down the search range.

[0054] The selection support process will be described with reference to FIG. First, the control unit 21 of the support device 20 executes a process of acquiring a sample image (step S11). Next, the control unit 21 of the support device 20 executes a process of narrowing down the search range (step S31). Specifically, the management unit 211 of the control unit 21 acquires the attributes specified by the user from the user device 10. Then, the extraction unit 212 extracts, from the item information storage unit 22, the item management information in which the specified attributes are recorded. Then, the control unit 21 of the support device 20 uses the extracted item management information to execute the process of extracting the designated number of similar images (step S12) and subsequent processes.

[0055] In the above embodiment, similar images are searched for based on sample images. Here, the search range may be narrowed down based on the selection history.

[0056] The process of narrowing down the search range will be described with reference to FIG. First, the control unit 21 of the support device 20 executes a selection history acquisition process (step S41). Specifically, the extraction unit 212 of the control unit 21 identifies the items selected by the user from the history information storage unit 23. As shown in FIG. 8, it is assumed that item candidates P2, P3, and P4 are selected in order for a sample image P1.

[0057] Next, the control unit 21 of the support device 20 executes a principal component analysis process of the history image (step S42). Specifically, the extraction unit 212 of the control unit 21 calculates image feature amounts of the identified item. Next, the extraction unit 212 performs principal component analysis of the calculated image feature amounts. Then, the extraction unit 212 identifies a component axis of each feature amount. Here, principal component analysis is performed on the image feature amounts of the item candidates P1 to P4.

[0058] Next, the control unit 21 of the support device 20 executes a process of evaluating the variation for each component axis (step S43). Specifically, the extraction unit 212 of the control unit 21 evaluates the dispersion state of the feature amount of the selection history image in the identified component axis. The dispersion state can be, for example, the dispersion (eigenvalue) or contribution rate of the data in the axis direction.

[0059] Next, the control unit 21 of the support device 20 performs a weighting process for the component axes according to the variations (step S44). Specifically, the extraction unit 212 of the control unit 21 compares the variations in the feature amounts of similar images for each component axis and weights the component axes. Here, a component axis with a large variation is weighted more heavily.

[0060] Next, the control unit 21 of the support device 20 executes a range narrowing process taking the weighting into consideration (step S45). Specifically, the extraction unit 212 of the control unit 21 calculates the distance of the similar range on the component axis according to the weighting.

[0061] As shown in FIG. 9, when the items included in the selection history are dispersed along a predetermined component axis AX1, weighting is performed on this component axis AX1 to widen the similarity range RG1. Furthermore, the method for selecting the representative item may be changed depending on the component axis AX1.

[0062] 10, it is assumed that clusters C21 to C23 have been generated in the similar image clustering process (step S13). In this case, for similar items included in cluster C22 that intersects with component axis AX1 with large variation, the nearest similar item 511 is identified as the representative item. On the other hand, for clusters C21 and C23 that do not intersect with the component axis with large variation, the similar item 512 that is close to the center of clusters C21 and C23 is identified as the representative item. This allows the range to be widened to extract similar items, since variations on the component axis indicate that the user's desires are not stable.

[0063] In the above embodiment, similar images are searched for based on sample images. Here, attributes may be narrowed down based on image feature amounts.

[0064] Next, the attribute feature axis generation process will be described with reference to Fig. 11(a). This process is performed prior to the item selection process (pre-processing). First, the control unit 21 of the support device 20 executes a calculation process of image feature amounts (step S51). Specifically, the management unit 211 of the control unit 21 calculates the image feature amounts of each item image.

[0065] Next, the control unit 21 of the support device 20 executes an attribute value acquisition process (step S52). Specifically, the management unit 211 of the control unit 21 acquires the attribute value of each item. For example, color, shape, pattern, etc. can be used as the attribute.

[0066] Next, the control unit 21 of the support device 20 executes a process of setting an attribute feature axis (step S53). Specifically, the management unit 211 of the control unit 21 associates the image feature amount with the attribute. Here, PLS-DA (Partial Least Squares Discriminant Analysis) is used as the discriminant analysis of the attribute feature axis. Note that the method is not limited to PLS-DA as long as it can perform discriminant analysis that associates the image feature amount with the attribute.

[0067] Next, the selection support process will be described with reference to FIG. First, the control unit 21 of the support device 20 executes a process for specifying attribute conditions by the user (step S61). Specifically, the extraction unit 212 of the control unit 21 acquires the attribute conditions specified by the user from the user device 10. For example, the attribute conditions specified by the user (same hue, color variations, etc.) are acquired.

[0068] Next, the control unit 21 of the support device 20 executes a narrowing-down process on the attribute feature axis according to the attribute conditions specified by the user (step S62). Specifically, the extraction unit 212 of the control unit 21 calculates image features according to the attributes on the attribute feature axis for the attribute conditions specified by the user. If a common attribute is desired with the preceding item, the variation in the attribute feature is adjusted so as to set a similarity range in a direction perpendicular to the attribute axis of this attribute. On the other hand, if a change in the attribute condition is desired with respect to the preceding item, the variation in the attribute feature is adjusted so as to set a similarity range in a direction along the attribute axis of this attribute.

[0069] As shown in FIG. 12(a), for example, when the same hue is desired for the color attribute (when a common attribute is desired), the extraction range is set in a direction perpendicular to the color component axis AX2. On the other hand, as shown in FIG. 12(b), when color variations (changes) are desired for the color attribute, the extraction range is set along the color component axis AX2. This allows the extraction range to be set using a component axis that evaluates image features based on item attributes. [Explanation of symbols]

[0070] 10...user device, 20...support device, 21...control unit, 211...management unit, 212...extraction unit, 213...candidate output unit, 22...item information storage unit, 23...history information storage unit.

Claims

1. An item search system comprising: a control unit connected to a user device; and an item information storage unit that stores a plurality of items, The control unit an item acquisition process for acquiring a search target item from the user device; calculating an image feature amount of the search target item, and identifying a specified number of similar items from the item information storage unit based on the image feature amount; generating a plurality of clusters by cluster analysis to classify the specified number of similar items into a plurality of groups; Identifying a positional relationship between each cluster and the search target item; changing a method for identifying a representative item according to the positional relationship; Identifying the representative item from among the plurality of similar items belonging to each cluster according to the positional relationship; executing an item search process for outputting the representative item for each group and outputting it to the user device; The item search system is characterized in that the representative item selected on the user device is used as a search target item and the item search process is repeated.

2. 2. The item search system according to claim 1, wherein the identification method involves identifying a similar item that is close to a center position of the cluster as the representative item.

3. 2. The item search system according to claim 1, wherein the method of specifying the representative item is to specify, in the cluster close to the search target item, a similar item closest to the search target item as the representative item.

4. An item search system comprising: a control unit connected to a user device; and an item information storage unit that records a plurality of items, The control unit an item acquisition process for acquiring a search target item from the user device; calculating an image feature amount of the search target item, and identifying a specified number of similar items from the item information storage unit based on the image feature amount; calculating a plurality of feature component axes with large variations in image feature component by principal component analysis for the specified number of similar items; Among the similar items grouped along the plurality of feature component axes, similar items having maximum and minimum values ​​are identified as representative items for each of the feature component axes; outputting the representative item and executing an item search process for outputting the representative item to the user device; The item search system is characterized in that the representative item selected on the user device is used as a search target item and the item search process is repeated.

5. An item search system comprising: a control unit connected to a user device; and an item information storage unit that records a plurality of items, The control unit an item acquisition process for acquiring a search target item from the user device; calculating an image feature amount of the search target item, and identifying a specified number of similar items from the item information storage unit based on the image feature amount; classifying the specified number of similar items into a plurality of groups; Identifying a representative item from among the similar items belonging to each group; executing an item search process for outputting the representative item for each group and outputting it to the user device; An item search system characterized in that the representative item selected on the user device is used as the search target item, the item search process is repeated, and the specified number is changed depending on the number of times the item search process is repeated.

6. An item search system comprising: a control unit connected to a user device; and an item information storage unit that records a plurality of items, The control unit an item acquisition process for acquiring a search target item from the user device; calculating image feature amounts of the search target item, calculating attribute feature axes by discriminant analysis using the attributes of the search target item and the image feature amounts, and adjusting a search range using the attribute feature axes in accordance with conditions related to attributes specified by a user on the user device, thereby identifying a specified number of similar items from the item information storage unit; classifying the specified number of similar items into a plurality of groups; Identifying a representative item from among the similar items belonging to each group; executing an item search process for outputting the representative item for each group and outputting it to the user device; The item search system is characterized in that the representative item selected on the user device is used as a search target item and the item search process is repeated.

7. An item search system comprising: a control unit connected to a user device; and an item information storage unit that records a plurality of items, The control unit an item acquisition process for acquiring a search target item from the user device; Calculating image features of the search target item; Get the item selection history, Principal component analysis is performed on the image features of the selected items, weighting the component axes calculated by the principal component analysis to narrow down the search range, and identifying a specified number of similar items from the item information storage unit using the image feature amounts; classifying the specified number of similar items into a plurality of groups; Identifying a representative item from among the similar items belonging to each group; executing an item search process for outputting the representative item for each group and outputting it to the user device; The item search system is characterized in that the representative item selected on the user device is used as a search target item and the item search process is repeated.

8. A method for searching for an item according to a user's preferences using an item search system including a control unit connected to a user device and an item information storage unit that stores a plurality of items, the method comprising: The control unit an item acquisition process for acquiring a search target item from the user device; calculating an image feature amount of the search target item, and identifying a specified number of similar items from the item information storage unit based on the image feature amount; generating a plurality of clusters by cluster analysis to classify the specified number of similar items into a plurality of groups; Identifying a positional relationship between each cluster and the search target item; changing a method for identifying a representative item according to the positional relationship; Identifying the representative item from among the plurality of similar items belonging to each cluster according to the positional relationship; executing an item search process for outputting the representative item for each group and outputting it to the user device; The item search method comprises repeating the item search process, with the representative item selected on the user device as the search target item.

9. A method for searching for items according to a user's preferences using an item search system including a control unit connected to a user device and an item information storage unit that records a plurality of items, comprising: The control unit an item acquisition process for acquiring a search target item from the user device; calculating an image feature amount of the search target item, and identifying a specified number of similar items from the item information storage unit based on the image feature amount; calculating a plurality of feature component axes with large variations in image feature component by principal component analysis for the specified number of similar items; Among the similar items grouped along the plurality of feature component axes, similar items having maximum and minimum values ​​are identified as representative items for each of the feature component axes; outputting the representative item and executing an item search process for outputting the representative item to the user device; The item search method comprises repeating the item search process, with the representative item selected on the user device as the search target item.

10. A method for searching for items according to a user's preferences using an item search system including a control unit connected to a user device and an item information storage unit that records a plurality of items, comprising: The control unit an item acquisition process for acquiring a search target item from the user device; calculating an image feature amount of the search target item, and identifying a specified number of similar items from the item information storage unit based on the image feature amount; classifying the specified number of similar items into a plurality of groups; Identifying a representative item from among the similar items belonging to each group; executing an item search process for outputting the representative item for each group and outputting it to the user device; An item search method characterized in that the representative item selected on the user device is used as a search target item, the item search process is repeated, and the specified number is changed depending on the number of times the item search process is repeated.

11. A method for searching for items according to a user's preferences using an item search system including a control unit connected to a user device and an item information storage unit that records a plurality of items, comprising: The control unit an item acquisition process for acquiring a search target item from the user device; calculating image feature amounts of the search target item, calculating attribute feature axes by discriminant analysis using the attributes of the search target item and the image feature amounts, and adjusting a search range using the attribute feature axes in accordance with conditions related to attributes specified by a user on the user device, thereby identifying a specified number of similar items from the item information storage unit; classifying the specified number of similar items into a plurality of groups; Identifying a representative item from among the similar items belonging to each group; executing an item search process for outputting the representative item for each group and outputting it to the user device; The item search method comprises repeating the item search process, with the representative item selected on the user device as the search target item.

12. A method for searching for items according to a user's preferences using an item search system including a control unit connected to a user device and an item information storage unit that records a plurality of items, comprising: The control unit an item acquisition process for acquiring a search target item from the user device; Calculating image features of the search target item; Get the item selection history, Principal component analysis is performed on the image features of the selected items, weighting the component axes calculated by the principal component analysis to narrow down the search range, and identifying a specified number of similar items from the item information storage unit using the image feature amounts; classifying the specified number of similar items into a plurality of groups; Identifying a representative item from among the similar items belonging to each group; executing an item search process for outputting the representative item for each group and outputting it to the user device; The item search method comprises repeating the item search process, with the representative item selected on the user device as the search target item.

13. An item search program that searches for an item according to a user's preferences using an item search system that includes a control unit connected to a user device and an item information storage unit that records a plurality of items, The control unit an item acquisition process for acquiring a search target item from the user device; calculating an image feature amount of the search target item, and identifying a specified number of similar items from the item information storage unit based on the image feature amount; generating a plurality of clusters by cluster analysis to classify the specified number of similar items into a plurality of groups; Identifying a positional relationship between each cluster and the search target item; changing a method for identifying a representative item according to the positional relationship; Identifying the representative item from among the plurality of similar items belonging to each cluster according to the positional relationship; executing an item search process for outputting the representative item for each group and outputting it to the user device; an item search program for causing the program to function as a means for repeating the item search process, with the representative item selected on the user device as a search target item;

14. An item search program for searching for items according to a user's preferences using an item search system having a control unit connected to a user device and an item information storage unit that records a plurality of items, The control unit an item acquisition process for acquiring a search target item from the user device; calculating an image feature amount of the search target item, and identifying a specified number of similar items from the item information storage unit based on the image feature amount; calculating a plurality of feature component axes with large variations in image feature component by principal component analysis for the specified number of similar items; Among the similar items grouped along the plurality of feature component axes, similar items having maximum and minimum values ​​are identified as representative items for each of the feature component axes; outputting the representative item and executing an item search process for outputting the representative item to the user device; an item search program for causing the program to function as a means for repeating the item search process, with the representative item selected on the user device as a search target item;

15. An item search program for searching for items according to a user's preferences using an item search system having a control unit connected to a user device and an item information storage unit that records a plurality of items, The control unit an item acquisition process for acquiring a search target item from the user device; calculating an image feature amount of the search target item, and identifying a specified number of similar items from the item information storage unit based on the image feature amount; classifying the specified number of similar items into a plurality of groups; Identifying a representative item from among the similar items belonging to each group; executing an item search process for outputting the representative item for each group and outputting it to the user device; An item search program that functions as a means for repeating the item search process using the representative item selected on the user device as the search target item and changing the specified number depending on the number of times the item search process is repeated.

16. An item search program for searching for items according to a user's preferences using an item search system including a control unit connected to a user device and an item information storage unit that records a plurality of items, The control unit an item acquisition process for acquiring a search target item from the user device; calculating image feature amounts of the search target item, calculating attribute feature axes by discriminant analysis using the attributes of the search target item and the image feature amounts, and adjusting a search range using the attribute feature axes in accordance with conditions related to attributes specified by a user on the user device, thereby identifying a specified number of similar items from the item information storage unit; classifying the specified number of similar items into a plurality of groups; Identifying a representative item from among the similar items belonging to each group; executing an item search process for outputting the representative item for each group and outputting it to the user device; an item search program for causing the program to function as a means for repeating the item search process, with the representative item selected on the user device as a search target item;

17. An item search program for searching for items according to a user's preferences using an item search system having a control unit connected to a user device and an item information storage unit that records a plurality of items, The control unit an item acquisition process for acquiring a search target item from the user device; Calculating image features of the search target item; Get the item selection history, Principal component analysis is performed on the image features of the selected items, weighting the component axes calculated by the principal component analysis to narrow down the search range, and identifying a specified number of similar items from the item information storage unit using the image feature amounts; classifying the specified number of similar items into a plurality of groups; Identifying a representative item from among the similar items belonging to each group; executing an item search process for outputting the representative item for each group and outputting it to the user device; an item search program for causing the program to function as a means for repeating the item search process, with the representative item selected on the user device as a search target item;

Images