JP2010039533A - Apparatus and method for image ranking, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To automatically rank images expected to have newsworthiness. <P>SOLUTION: An image ranking apparatus includes: a crawling engine 11 which obtains images and a site URL with the images posted thereon from the Internet 30, and registers first image information, where the hash values of the collected images are associated with the site URL with the images posted thereon, etc., in a first image information table in a DB 15; an image hash comparing device 14 which compares the hash values of the collected images with hash values preserved in a second image information table, and, when the hash value being the same as the hash value is preserved, adds one to a counting number related to the hash value in the second image information table, when the hash value being the same as the hash value is not preserved, generates the second image information table where the hash value is related to the counting number one; and an image ranking display generating device 21 which performs image ranking, based on the second image information table. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an image ranking apparatus, an image ranking method, and a program, and more particularly to an image ranking apparatus, an image ranking method, and a program capable of ranking images posted on the Internet.

  The following is disclosed as an invention for automatically generating a ranking based on information on the Internet.

  In Patent Literature 1, when a keyword is input, a search is performed for a web page including the keyword, a search keyword log is created, and the search keyword log is aggregated for each search keyword to determine the ranking of the search keyword. A keyword presentation method is disclosed.

According to Patent Document 2, a blog article is acquired based on an input keyword, a relationship between the blog site and the blog article is extracted for each keyword, and the score of the blog site and the keyword is calculated based on the result. An information image ranking device that ranks blog sites and keywords based on the score obtained is disclosed.
JP 2004-206517 A JP 2007-334590 A

  However, in the inventions described in Patent Documents 1 and 2, it is possible to know keyword rankings, that is, topical keywords, but there are the following problems when applied to topical images.

  For example, a method of inputting a keyword, ranking images tagged with the keyword, and displaying a list of the results can be considered. However, in this case, the search condition depends on the keyword information, and there is a problem that an appropriate keyword must be input. Further, since the search is performed based on the keyword, information on the image itself is not used. That is, it is difficult to say that the search is performed based on the topicality of the image itself.

  Another possible method is to input an image instead of a keyword and search for the URL and the number of pages on which the same image as the image input on the Internet is posted. When the exact same image is posted on multiple pages on the Internet, it is considered that the image has some topicality and popularity, and the more the number is posted, the more topical the image. This is because it is thought to be possessed. However, since it is impossible to check the number of the same images without inputting the images as keys, it is not suitable for searching for topical images.

  The present invention has been made in view of such circumstances, and an image ranking apparatus, an image ranking method, and a program capable of automatically ranking and displaying images considered to be topical without inputting keywords or the like The purpose is to provide.

  The image ranking apparatus according to claim 1, a crawling unit that collects images from the Internet, a thumbnail image generating unit that generates thumbnail images of images acquired by the crawling unit, and an image collected by the crawling unit Identification information calculating means for calculating the identification information of the image from, a thumbnail image generated from the image for each image collected by the crawling means, and identification information calculated from the image are associated with each other. First image information registration means for registering one image information in the first image information table, based on the first image information registered in the first image information table, identification information of the image, A second image information table creation method for creating a second image information table in which the number of registrations for each identification information is associated Extracting the identification information from the second image information table in order of the number of registrations, extracting the thumbnail image of the image corresponding to the extracted identification information from the first image information table, And a ranking display screen output means for outputting a ranking display screen showing the order of the images with the most registered numbers.

  According to the image ranking apparatus of the first aspect, an image is collected from the Internet, a thumbnail image is generated from the image, and identification information (hash value, feature amount, etc.) of the image is calculated. For each collected image, the first image information in which the thumbnail image generated from the image and the identification information calculated from the image are associated is registered in the first image information table, and the registered first image information is stored. Based on the image information, a second image information table is created in which the identification information of the image is associated with the number of registrations for each identification information. The identification information is extracted from the second image information table in descending order of the number of registrations, the thumbnail images of the images corresponding to the identification information extracted from the first image information table are extracted, and the ranking including the extracted thumbnail images is included. A ranking display screen showing the order of images with a large number of registrations is output. Accordingly, it is possible to automatically display rankings of images that are frequently posted on the Internet, that is, images that are considered to be topical.

  The image ranking device according to claim 2 is the image ranking device according to claim 1, wherein the crawling means includes, together with the image, a site URL where the image is posted, a web page URL, and an image URL. Collecting at least one URL, the first image information registration means includes a URL collected corresponding to the image, a thumbnail image generated from the image, and identification information calculated from the image. The associated first image information is registered in a first image information table, and the ranking display screen output means extracts thumbnail images and URLs of images with a large number of registrations from the first image information table, A ranking display screen including the extracted thumbnail image and the link button of the URL is output.

  According to the image ranking apparatus of claim 2, at least one URL of a site URL, a web page URL, and an image URL on which the image is posted is collected together with the image, and is collected corresponding to the image. The first image information in which the URL, the thumbnail image generated from the image, and the identification information calculated from the image are associated is registered in the first image information table. A thumbnail image and URL of an image with a large number of registrations are extracted from the first image information table, and a ranking display screen including the extracted thumbnail image and the link button of the URL is output. Thereby, the user can browse the page on which the ranked image is posted without performing a search.

  The image ranking apparatus according to claim 3 determines whether or not the identification information calculated by the identification information calculation means is stored in the first image information in the image ranking apparatus according to claim 1 or 2. And when the determination unit determines that the identification information is stored in the first image information, a number obtained by adding 1 to the current count number is set as a new count number, When the determination means determines that the identification information is not stored in the first image information, the second image information table creation means includes a count number calculation means that sets the count number to 1. Is characterized in that the count number calculated by the count number calculating means is the registered number.

  According to the image ranking apparatus of the third aspect, it is determined whether or not the identification information calculated from the images collected from the Internet is stored in the first image information table. Then, the number obtained by adding 1 to the current count number is set as a new count number, and when not stored, the count number is set as 1, and the second image information is created by using the count number as the registered number. Accordingly, the number of identification information included in the first image information can be calculated as the count number.

  The image ranking device according to claim 4 is the image ranking device according to claim 3, wherein the crawling means collects a URL of a site where the image is posted together with the image, and registers the first image information. The means registers, in the first image information table, first image information in which a thumbnail image generated from the image, identification information calculated from the image, and a site URL on which the image is posted are associated with each other. The determination means determines whether or not a plurality of combinations in which the same site URL is associated with the same identification information is stored in the first image information, and the count number calculation means determines the determination If it is determined by the means that a plurality of combinations of the same identification information and the same site URL are stored in the first image information, the plurality of storages are stored. And calculates the count number a combination of the same site URL and the same identification information as one.

  According to the image ranking apparatus of claim 4, when a plurality of combinations of the same identification information and the same site URL are stored in the first image information, a plurality of the same identification information stored. The number of counts is calculated assuming that the combination with the same site URL is 1. As a result, it is possible to prevent the case where the same image is reused in a huge site, or to prevent multiple counting due to the difference in page notation, and to calculate an accurate count. Therefore, accurate ranking can be performed.

  The image ranking device according to claim 5 is the image ranking device according to any one of claims 1 to 4, wherein the crawling means acquires date and time information when the image is collected together with the image, and The first image information registering means stores the first image information in which the thumbnail image generated from the image, the identification information calculated from the image, and the date / time information when the image is collected are associated with each other. The second image information table creation means creates the second image information table for the images collected by the crawling means within a predetermined period based on the date / time information. It is characterized by that.

  According to the image ranking device of claim 5, the date and time information when the image is collected is acquired together with the image, the thumbnail image generated from the image, the identification information calculated from the image, and the image are collected. The first image information associated with the date / time information is registered in the first image information table. Based on this date and time information, a second image information table is created for the images collected by the crawling means within a predetermined period. Thereby, ranking for every predetermined period can be performed automatically.

  The image ranking apparatus according to claim 6 is the image ranking apparatus according to any one of claims 1 to 5, wherein the ranking display output means includes a link button to the ranking display screen, together with the ranking display screen, A module including a thumbnail image included in the ranking display screen is output.

  According to the image ranking apparatus of the sixth aspect, the ranking display is output as a module on the Internet. By publishing this module on the Internet and having it posted on other people's blogs, you can increase the number of accesses to your site and increase advertising revenue.

  The image ranking device according to claim 7 is the image ranking device according to any one of claims 1 to 6, wherein the identification information is a hash value calculated by applying a hash function to the image. It is characterized by being. As a result, the same image can be reliably recognized as the same image.

  An image ranking apparatus according to an eighth aspect is the image ranking apparatus according to any one of the first to sixth aspects, wherein the identification information is a feature amount indicating a characteristic unique to the image. As a result, similar images that have been enlarged or reduced can be recognized as the same image. Therefore, ranking with higher accuracy can be performed.

  The image ranking method according to claim 9, a step of collecting an image from the Internet, a step of generating a thumbnail image of the acquired image, and a step of calculating identification information of the image from the collected image Registering, in the first image information table, first image information in which thumbnail images generated from the collected images are associated with identification information calculated from the images; And creating a second image information table in which the identification information of the image and the registration number for each identification information are associated with each other based on the first image information registered in the first image information table. And extracting the identification information from the second image information table in the descending order of the number of registrations, and extracting the identification information extracted from the first image information table. Extracting a thumbnail image of an image corresponding to the ranking image, and outputting a ranking display screen including the extracted thumbnail image, the ranking display screen showing the order of the images having a large number of registrations. To do.

  A program according to a tenth aspect causes an arithmetic device to execute the image ranking method according to the ninth aspect.

  According to the present invention, it is possible to automatically display rankings of images considered to be topical without inputting keywords or the like.

<First Embodiment>
FIG. 1 is a schematic diagram of the overall structure of an image ranking apparatus 1 according to the first embodiment. The image ranking device 1 is mainly composed of a crawling unit 10 and a ranking unit 20. The crawling unit 10 mainly includes a crawling engine 11, an image hash calculation device 12, a thumbnail image generation unit 13, an image hash comparison device 14, and a database (DB) 15. The ranking unit 20 includes an image ranking. The display generation device 21 and the display device 22 are included. The Internet 30 is connected to the crawling engine 11 and the image ranking display generation device 21.

  The crawling engine 11 connects to a web page on the Internet 30 and acquires the web page URL and the site URL. The crawling engine 11 traces link information described in HTML (Hyper Text Markup Language) or the like, and acquires an image URL posted on a web page and its image data. The image data acquired by the crawling engine 11 is output to the image hash calculation device 12 and the thumbnail image generation means 13.

  The image hash calculation device 12 calculates a hash value of an image output from the crawling engine 11 using a hash function such as MD5 or SHA-1. The hash value calculated by the image hash calculation device 12 is input to the crawling engine 11.

  The thumbnail image generation unit 13 generates a thumbnail image of the image output from the crawling engine 11. The thumbnail image generated by the thumbnail image generating means 13 is input to the crawling engine 11.

  The crawling engine 11 outputs the hash value of the image output from the image hash calculation device 12 from the thumbnail image generation unit 13 based on the web page URL, the site URL, the image URL, and the image from which the image was acquired. First image information associated with a thumbnail image or the like is registered in a first image information table in the DB 15. The crawling engine 11 circulates the Internet 30 automatically and recursively performs this process, thereby continuously increasing the first image information table.

  An example of the data structure of the first image information table is shown in FIG. The first image information table includes a web page URL on which an image is posted, an image URL posted on the web page, a site URL of the web page, a hash value of the image, and a thumbnail of the image. First image information associated with an image is stored. The hash value is an index for detecting the identity of images, and the same hash value is calculated from the same image. In the case illustrated in FIG. 2, the same hash value “123456789” is stored in the first row, the fourth row, and the seventh row, indicating that these three are the same image.

  The image hash comparison device 14 compares the hash value of the image output from the image hash calculation device 12 with the first image information table stored in the DB 15, and the hash value of the image is stored in the first image information table. Is calculated (hereinafter referred to as the count number). The process of calculating the count number by the image hash comparison device 14 will be described in detail later. The count number calculated by the image hash comparison device 14 is input to the crawling engine 11.

  The crawling engine 11 creates a second image information table in which the hash value of the image and the count number are associated with each other based on the first image information, and stores the second image information table in the DB 15.

  The DB 15 stores a first image information table as shown in FIG. 2, a second image information table in which the hash value of the image as shown in FIG. Also, the date and time of crawling and detailed information of the image acquired at that time (image file name, vertical / horizontal width, file size, acquisition date / time, etc.) are stored in the DB 15 in association with each other.

  The image ranking display generation device 21 generates screen data for image ranking display based on the first image information table and the second image information table stored in the DB 15. The image ranking display screen data generated by the image ranking display generation device 21 is output to the display device 22 or the Internet 30. The image ranking display screen generated by the image ranking display generation device 21 will be described in detail later.

  The display device 22 is a liquid crystal display capable of color display, and displays the image ranking display screen output from the image ranking display generation device 21.

  The operation of the image ranking apparatus 1 configured as described above will be described. First, the operation of the crawling unit will be described. FIG. 4 is a flowchart showing a flow of processing performed by the crawling unit 10.

  The crawling engine 11 acquires an image on the Internet 30, a web page URL where the image is posted, a site URL, and an image URL (step S1). The image acquired in step S1 is input to the image hash calculation device 12 and the thumbnail image generation means 13.

  The image hash calculation device 12 calculates the hash value of the image acquired in step S1 (step S2), and the thumbnail image generation unit 13 generates a thumbnail image of the image acquired in step S1 (step S3). The hash value calculated in step S2 and the thumbnail image generated in step S3 are input to the crawling engine 11.

  The crawling engine 11 includes first image information obtained by combining the web page URL, site URL, and image URL acquired in step S1, the hash value calculated in step S2, and the thumbnail image generated in step S3. Is registered in the first image information table of the DB 15 (step S4).

  The crawling engine 11 determines whether crawling has ended, that is, whether all images on the Internet 30 have been collected (step S5). If crawling has ended (YES in step S5), the process ends.

  If the crawling has not ended (NO in step S5), the crawling engine 11 determines that the site URL acquired at the same time as the hash value calculated in step S2 and the image on which the hash value is based, and the first image The hash value and the site URL stored in the information table are compared (step S6), and the first value obtained by combining the hash value calculated in step S2 and the site URL acquired at the same time as the image on which the hash value is based is combined. It is determined whether or not the first image information in which the same hash value and site URL as the image information are combined is already stored in the first image information table (step S7).

  The first image information having the same combination as the first image information obtained by combining the hash value calculated in step S2 and the site URL acquired at the same time as the image on which the hash value is based is already stored in the first image information table. If YES (YES in step S7), the hash value counting process (steps S8 to S11) is not performed, the process returns to step S1, and the crawling process (steps S1 to S7) is performed again. Here, not only the hash value but also the site URL is compared in the case where the same image is reused in a huge site such as Yahoo, or http: // www. xxx. jp / and http: // www. xxx. jp / index. This is to prevent multiple counting due to differences in page notation such as html.

  First image information having the same combination as the first image information obtained by combining the hash value calculated in step S2 and the site URL acquired simultaneously with the image that is the basis of the hash value is stored in the first image information table. If not (NO in step S7), the process proceeds to the hash value counting process (steps S8 to S11).

  The image hash comparison device 14 compares the hash value calculated in step S2 with the hash value stored in the second image information table (step S8), and the hash value same as the hash value calculated in step S2 is obtained. It is determined whether it is stored in the second image information table (step S9).

  If the same hash value as the hash value calculated in step S2 is stored in the second image information table (YES in step S9), the image hash comparison device 14 calculates the hash value in the second image information table. 1 is added to the associated count number (step S10). For example, when the second image information table shown in FIG. 3 is stored in the DB 15 and the hash value calculated in step S2 is 12345789, the image hash comparison device 14 stores the second image information table in the second image information table. 1 is added to 950 which is the current count number, and the count number is changed to 951.

  When the same hash value as the hash value calculated in step S2 is not stored in the second image information table (NO in step S9), the crawling engine 11 associates the hash value with the count number 1 and stores the second hash value. Save in the image information table (step S11). For example, if the second image information table shown in FIG. 3 is stored in the DB 15 and the hash value calculated in step S2 is 111111111, the hash value 111111111 is stored in the second image information table. Therefore, the hash value 111111111 is registered in the second image information table with the count number being 1.

  As a result, the hash value counting process (steps S8 to S11) ends, and the number of hash values stored in the first image information table is calculated as the count number. Returning to step S1, the crawling process (steps S1 to S7) is performed again.

  Thereby, the crawling unit 10 can continuously update the first image information table and the second image information table, and can always keep the first image information table and the second image information table in the latest state.

  Next, the operation of the ranking unit 20 will be described. 5 to 8 are image ranking display screens generated by the image ranking display generation device 21. FIG.

  The image ranking display generation device 21 performs ranking in descending order of the number of points with the count number of the second image information table as a point. For example, in the case shown in FIG. 3, the image ranking display generation device 21 ranks the image with a hash value of 123456789 as 1st ranking with 950 points, the image with the hash value of 5656656565 as 2nd ranking with 892 points, and the hash value Ranking is performed in such a manner that the image of 7878787787 is ranked third with 777 points. The point is not limited to the count number, and a numerical value calculated based on the count number may be used.

  FIG. 5 is an example of an image ranking display screen. When a user inputs an image ranking display instruction to the image ranking device 1 using an input device (not shown), a screen (top page) that is first displayed on the display device 22. It is. The image ranking display generation device 21 generates a screen displaying thumbnail images, the number of points, etc. in the ranking order as a top page. As thumbnail images displayed on the top page, thumbnail images stored in the first image information table are used. Note that the size of the thumbnail image to be displayed may be changed according to the order as shown in FIG.

  Also, the top page shown in FIG. 5 is provided with a soft button or the like with a link to another page for each ranking. For example, a soft button written as “view duplicate images” has a link to the web page URL list display screen shown in FIG. 6, and thumbnail images go to the image detailed information screen shown in FIG. The link is stretched. When the user clicks a soft button or a thumbnail image, the page transitions to the page of FIG. 6 or FIG.

  FIG. 6 shows an example of a web page URL list display screen, in which a list of URLs of web pages including the image ranked first is displayed. The screen shown in FIG. 6 is a screen displayed when the user clicks a soft button “view duplicate images” displayed adjacent to the image ranked first in the top page shown in FIG. It is. The image ranking display generation device 21 acquires a web page URL from the first image information table, and generates a screen in which the acquired web page URLs are arranged together with, for example, 50-sound order and ABC order and displayed together with thumbnail images and points. In addition, although the example which displays 10 web page URLs every time was shown in FIG. 6, a display form is not limited to this.

  Further, the web page URL list display screen shown in FIG. 6 is provided with a soft button or the like with a link to another page. For example, in the soft button written as “Previous 10”, the previous 10 to 20 of 950 currently displayed, that is, 1 to 10 of 950 are displayed. There is a link to the page, and the soft button labeled "Next 10" shows 10 out of 950 currently displayed, that is, 10 after -20, that is, out of 950 A link to a page displaying 20 to 30 is attached, and a link to the top page shown in FIG. 5 is attached to the characters “return to ranking list”. Further, the thumbnail image has a link to the image detailed information screen shown in FIG.

  FIG. 7 shows an example of the image detailed information screen, which displays detailed information of the image ranked first in the image ranking. The screen shown in FIG. 7 is a screen displayed when the user clicks the thumbnail image of the image ranked first in the top page shown in FIG. The image ranking display generation device 21 acquires detailed image information, information such as the date and time of crawling from the DB 15, acquires the web page URL on which the image was posted from the first image information table, and these are thumbnail images. , Generate a screen displayed with points.

  Further, the image detailed information screen shown in FIG. 7 is provided with a soft button or the like with a link to another page. For example, a soft button written “view duplicate images” has a link to the web page URL list display screen shown in FIG. 6, and a soft button written “previous” contains There is a link to the detailed image information screen that is one rank higher than the ranking of the displayed image, and the soft button labeled “Next” is 1 from the ranking of the currently displayed image. There is a link to the image detail information screen of the next order. The thumbnail image has a link to the image URL.

  FIG. 8 is an example when the image ranking display is a module, for example, a blog part, (a) is an example of a blog with a blog part pasted, and (b) is pasted on the blog of (a). It is the figure which expanded the attached blog part. Blog parts are systems that display information on other sites in a small frame to improve the functionality and design of the blog, and blog parts that contain information about your site are published on the Internet. However, it is intended to increase the number of accesses to its own site and increase advertising revenue by having it posted on other people's blogs. FIG. 8 shows an example of providing a blog part as a module, but the module is not limited to a blog part but may be a banner or the like.

  The blog part shown in FIG. 8B has a width of about 180 pixels, and the image ranking display generation device 21 displays a thumbnail image, a ranking, and a point of an image with a desired ranking. Create parts using Flash, GIF, etc. In the blog part, you may make it display only the image of the desired ranking, the ranking, and the point, or, for example, images ranked 1 to 10 in a slide show format that transitions to an introduction screen of a different order after a certain period of time, Rankings and points may be displayed in order.

  In addition, the blog part shown in FIG. 8 is provided with a soft button or the like with a link to another page. For example, the soft button labeled “Go to ranking site!” Has a link to the top page shown in FIG. 5, and the soft button labeled “Previous” is currently displayed. There is a link to the introduction screen that is one rank higher than the ranking of the current image, and the soft button labeled “Next” has a rank one lower than the ranking of the currently displayed image. There is a link to the introduction screen. Further, the thumbnail image has a link to the top page shown in FIG.

  The image ranking display generation device 21 outputs the generated top page and image detailed information screen to the display device 22 or the Internet 30. In addition, the image ranking display generation device 21 outputs the generated module to the Internet 30.

  According to the present embodiment, by acquiring an image from the Internet by crawling and calculating the number of counts of the image, an image that is frequently posted on the Internet, that is, an image that is considered to be topical is automatically acquired. Ranking can be displayed.

  In addition, according to the present embodiment, the image on the Internet is crawled using the hash value of the image, so that the same image can be reliably searched. Further, since a hash value is calculated from the acquired image and a search is performed using the hash value, the user does not need to input a keyword or the like, and usability can be improved.

  In the present embodiment, the date and time when crawling is performed in the DB 15 and the detailed information of the image acquired at that time are stored in association with each other. However, the date and time when crawling is performed is associated with the hash value and the first image. It may be stored in an information table. In this case, by comparing the hash values for each desired period with the image hash comparison device 14, a plurality of second image information tables storing count numbers for each desired period can be created. Therefore, the image ranking display generation device 21 can generate an image ranking display screen for each desired period.

  Further, according to the present embodiment, the count number is calculated every time crawling is performed. However, the count number may be calculated as needed instead of being calculated every time crawling is performed. For example, the count number may be calculated when a certain period has elapsed, or the count number may be calculated when a certain number of first image information is registered in the first image information table.

<Second Embodiment>
In the first embodiment, an image on the Internet is crawled using a hash value, but is calculated from image data instead of a hash value as search information for searching a URL where an image to be searched is posted. Various possible features can be used.

  In the second embodiment, a feature amount indicating an image-specific feature such as an image pattern, color, and brightness is calculated, and the image is crawled using the feature amount. Hereinafter, the image ranking apparatus 2 according to the second embodiment will be described. The same parts as those in the first embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

  FIG. 9 is a schematic diagram of the overall structure of the image ranking apparatus 2 that uses the feature amount to crawling an image on the Internet. The image ranking device 2 mainly includes a crawling unit 10 ′ and a ranking unit 20. The crawling unit 10 ′ is mainly composed of a crawling engine 11, an image feature amount calculation device 16, a thumbnail image generation unit 13, a database (DB) 15, and an image feature amount comparison device 17.

  The image feature amount calculation device 16 calculates the feature amount of the image input from the crawling engine 11. The feature amount is a value representing image-specific features such as image design, color, and brightness, and is given as, for example, a multidimensional vector. For example, the image feature amount calculation device 16 has an upper left area (area 1), a lower left area (area 2), an upper right area (area 3), and a lower right area (area 4) when the image is divided into four. ) And the average value of the colors (for example, Y component) of the five areas of the central area (area 5) of the image, and the colors of the areas 1, 2, 3, 4, and 5 are calculated. A five-dimensional vector using the average value is calculated as a feature amount.

  As a result, if images (similar images) having the same design, color, brightness, etc., or different in design, color, brightness, etc. are indistinguishable from the user's perspective, they are the same or substantially the same (± Feature quantity (with a slight error of about one digit) can be calculated. That is, even if processing such as enlargement / reduction of an image or change of a storage format (for example, jpeg format to bmp format) is performed, the same or substantially the same if it cannot be distinguished from the user A feature amount can be calculated.

  The image feature amount comparison device 17 compares the feature amount output from the image feature amount calculation device 16 with the first image information table stored in the DB 15 and calculates the count number. The process of calculating the count number by the image feature amount comparison device 17 will be described in detail later.

  The operation of the image ranking device 2 configured as described above will be described. Since the operation of the ranking unit 20 is the same as that of the first embodiment, the operation of the crawling unit 10 'will be described. FIG. 10 is a flowchart showing a flow of processing performed by the crawling unit 10 '.

  The crawling engine 11 acquires an image on the Internet 30, a web page URL where the image is posted, a site URL, and an image URL (step S1).

  The image feature quantity calculation device 16 calculates the feature quantity of the image acquired in step S1 (step S12), and the thumbnail image generation means 13 generates a thumbnail image of the image acquired in step S1 (step S3). . The feature amount calculated in step S12 and the thumbnail image generated in step S3 are input to the crawling engine 11.

  The crawling engine 11 sets the first page of the DB 15 by combining the web page URL, site URL, and image URL acquired in step S1, the feature amount calculated in step S12, and the thumbnail image generated in step S3. Register in the image information table (step S13). As a result, the web page URL, the image URL, the site URL, the feature amount, and the thumbnail image of the image are stored in the first image information table in association with each other.

  The crawling engine 11 determines whether crawling has been completed (step S5). If crawling has ended (YES in step S5), the process ends.

  If crawling has not ended (NO in step S5), the crawling engine 11 uses the feature amount and site URL calculated in step S12 and the feature amount and site URL stored in the first image information table. The comparison is performed (step S14), and it is determined whether or not the first image information in which the same feature amount and site URL as the feature amount and site URL calculated in step S12 are combined is already stored in the first image information table (step S14). Step S15).

  If the first image information in which the same feature quantity and site URL as the feature quantity and site URL calculated in step S12 are combined is already stored in the first image information table (YES in step S15), the feature quantity The counting process (steps S16 to S11) is not performed, and the process returns to step S1 and the crawling process (steps S1 to S15) is performed again. Here, not only the feature amount but also the site URL is compared, as in the first embodiment, a case where the same image is reused in a huge site or a difference in page notation is counted. This is to prevent things.

  If the first image information in which the same feature amount and site URL as the feature amount and site URL calculated in step S12 are combined is not stored in the first image information table (NO in step S15), the feature amount The process proceeds to the count process (steps S16 to S19).

  The image feature quantity comparison device 17 compares the feature quantity calculated in step S12 with the feature quantity stored in the second image information table (step S16), and the same feature quantity as the feature quantity calculated in step S12. Is stored in the second image information table (step S17). At this time, a slight error of about ± 1 digit is allowed, and approximately the same feature amount may be handled as the same feature amount. As a result, similar images that have been processed such as enlargement / reduction, but are indistinguishable from the original image when viewed from the user, can be counted as the same image.

  When the same feature quantity as that calculated in step S12 is stored in the second image information table (YES in step S17), the image feature quantity comparison device 17 performs the feature quantity in the second image information table. 1 is added to the count number associated with (step S18).

  If the same feature quantity calculated in step S12 is not stored in the second image information table (NO in step S17), the image feature quantity comparison device 17 associates the feature quantity with the count number 1. Is stored in the second image information table (step S19).

  When the feature amount counting process (steps S16 to S19) ends, the process returns to step S1 and the crawling process (steps S1 to S15) is performed again.

  Thereby, the crawling unit 10 ′ can continuously update the first image information table and the second image information table, and can always keep the first image information table and the second image information table in the latest state.

  According to the present embodiment, not only a completely matched image, but also an image in which the original image is enlarged or reduced, an image re-saved at a reduced resolution, an image saved in a different format, and the like are matched. Since it can be distinguished from an image, the accuracy of image ranking can be increased.

  By the way, in the present invention, since images are automatically collected and ranked, harmful images such as adult images may be displayed on the image ranking display screen. As a technique for avoiding such a concern, the following method can be considered.

(1) Filtering software format (client side)
This is a method of displaying a screen for image ranking display after filtering software is downloaded in advance to the image ranking apparatus. Thereby, it is possible to prevent browsing of only the harmful image portions on the image ranking display screens as shown in FIGS.

(2) Filtering software method (provided by provider)
In this method, the crawling engine 11 performs crawling via a provider corresponding to harmful image elimination. Thereby, the first image information table and the second image information table excluding the harmful image can be generated. Accordingly, the ranking without the harmful images is displayed on the image ranking display screen.

(3) White list method (server side)
This is a method of previously limiting the hosts to be crawled. A host to be crawled is registered in the crawling engine 11, and the crawling engine 11 performs crawling only on the registered host. Thereby, the first image information table and the second image information table excluding the harmful image can be generated. Accordingly, the ranking without the harmful images is displayed on the image ranking display screen.

(4) Blacklist method (server side)
This is a method to remove harmful sites from crawling according to the black list published by the police. A site to be excluded from the crawling target is registered in the crawling engine 11, and the crawling engine 11 does not perform crawling on the registered site. Thereby, the first image information table and the second image information table excluding the harmful image can be generated. Accordingly, the ranking without the harmful images is displayed on the image ranking display screen.

  The present invention can be applied not only to still images but also to moving images, and also to various contents other than images such as music.

  The present invention may be provided as a device including a crawling unit and a ranking unit, or may be provided as a system including two or more devices. The apparatus may be a PC or a mobile terminal. Further, the present invention is not limited to the device, and can be provided as a program applied to the device.

It is the schematic of the image ranking apparatus 1 of the 1st Embodiment of this invention. It is an example of the data structure of the 1st image information table of the image ranking apparatus 1. FIG. It is an example of the data structure of the 2nd image information table of the image ranking apparatus. 4 is a flowchart showing a process flow of a crawling unit 10 of the image ranking device 1. 3 is an example of an image ranking display screen of the image ranking device 1. 3 is an example of an image ranking display screen of the image ranking device 1. 3 is an example of an image ranking display screen of the image ranking device 1. 3 is an example of an image ranking display screen of the image ranking device 1. It is the schematic of the image ranking apparatus 2 of the 2nd Embodiment of this invention. 4 is a flowchart illustrating a processing flow of a crawling unit 10 ′ of the image ranking device 2.

Explanation of symbols

1, 2: Image ranking device, 10, 10 ′: Crawling unit, 11: Crawling engine, 12: Image hash calculation device, 13: Thumbnail image generation means, 14: Image hash comparison device, 15: Database (DB), 20: Ranking unit, 21: Image ranking display generation device, 22: Display device, 30: Internet

Claims (10)

A crawling means for collecting images from the Internet;
Thumbnail image generating means for generating a thumbnail image of the image acquired by the crawling means;
Identification information calculating means for calculating identification information of the image from the image collected by the crawling means;
First image information in which a thumbnail image generated from the image for each image collected by the crawling means and identification information calculated from the image are associated is registered in a first image information table. 1 image information registration means;
Based on the first image information registered in the first image information table, a second image information table is created in which the identification information of the image is associated with the registration number for each identification information. Image information table creation means,
The identification information is extracted from the second image information table in the descending order of the number of registrations, the thumbnail image of the image corresponding to the extracted identification information is extracted from the first image information table, and the extracted thumbnails are extracted. A ranking display screen output means for outputting a ranking display screen that includes an image and outputs a ranking display screen indicating an order of the images having a large number of registrations;
An image ranking apparatus comprising: The crawling means collects, together with the image, at least one URL of a site URL, a web page URL, and an image URL where the image is posted,
The first image information registration unit includes first image information in which a URL collected corresponding to the image, a thumbnail image generated from the image, and identification information calculated from the image are associated with each other. Is registered in the first image information table,
The ranking display screen output means extracts thumbnail images and URLs of the registered images from the first image information table, and outputs a ranking display screen including the extracted thumbnail images and URL link buttons. The image ranking apparatus according to claim 1, wherein: Determination means for determining whether or not the identification information calculated by the identification information calculation means is stored in the first image information;
When the determination means determines that the identification information is stored in the first image information, a number obtained by adding 1 to the current count number is set as a new count number, and the determination means determines the identification information. If it is determined that the information is not stored in the first image information, a count number calculating means for setting the count number to 1,
3. The image ranking apparatus according to claim 1, wherein the second image information table creation unit uses the count number calculated by the count number calculation unit as the registration number. 4. The crawling means collects a site URL where the image is posted together with the image,
The first image information registration means obtains first image information in which a thumbnail image generated from the image, identification information calculated from the image, and a site URL where the image is posted are associated with each other. Registered in the image information table of 1,
The determination means determines whether or not a plurality of combinations in which the same site URL is associated with the same identification information are stored in the first image information,
When the determination unit determines that a plurality of combinations of the same identification information and the same site URL are stored in the first image information, the count number calculation unit stores the plurality of the same stored information. The image ranking apparatus according to claim 3, wherein the count number is calculated by setting a combination of the identification information of the URL and the same site URL as 1. The crawling means acquires the date and time information when the image was collected together with the image,
The first image information registration means obtains first image information in which a thumbnail image generated from the image, identification information calculated from the image, and date / time information when the image is collected are associated with each other. Registered in the image information table of 1,
The second image information table creating unit creates the second image information table for images collected by the crawling unit within a predetermined period based on the date and time information. The image ranking apparatus according to any one of 1 to 4.   6. The ranking display output means outputs a module including a link button to the ranking display screen and a thumbnail image included in the ranking display screen together with the ranking display screen. The image ranking apparatus according to any one of the above.   The image ranking apparatus according to claim 1, wherein the identification information is a hash value calculated by applying a hash function to the image.   The image ranking apparatus according to claim 1, wherein the identification information is a feature amount indicating a characteristic unique to the image. Collecting images from the internet;
Generating a thumbnail image of the acquired image;
Calculating identification information of the image from the collected image;
Registering, in the first image information table, first image information in which a thumbnail image generated from the image for each collected image is associated with identification information calculated from the image;
Creating a second image information table in which the identification information of the image and the number of registrations for each identification information are associated with each other based on the first image information registered in the first image information table; ,
The identification information is extracted from the second image information table in the descending order of the number of registrations, the thumbnail image of the image corresponding to the extracted identification information is extracted from the first image information table, and the extracted thumbnails are extracted. A ranking display screen including images, the step of outputting a ranking display screen indicating the order of the images with a large number of registrations;
An image ranking method comprising:   A program causing an arithmetic device to execute the image ranking method according to claim 9.

Images