JP2009282883A - Image retrieval system, crawling device, and image retrieval device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To allow a user to easily know where a desired image is posted on the Internet. <P>SOLUTION: A crawling engine 11 is connected to a web page on the Internet 40 so as to acquire a URL of the web page, a URL of an image posted on the web page, and its image data. A hash value of the collected image is calculated by an image hash calculating device 12 so as to store the hash value in an image DB 30 by making it relate to the URL of the web page from which the image is acquired, and the URL or the like of the image. A retrieval device 23 checks whether or not a hash value coincident with the same hash value as that outputted from the image hash calculating device 22 is included in the image DB 30 and acquires the URL of the web page made to relate to the hash value and the URL of the image from the image DB 30 so as to display them on a retrieval result display device 24. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an image search system, a crawling device, and an image search device, and more particularly to an image search system, a crawling device, and an image search device that can search for a desired image on the Internet.

  In Patent Literature 1, keywords are extracted from the contents of search conditions using natural sentences, information similar to the keywords is extracted from a database in which information collected from the Internet is registered, and the results are in descending order of similarity to the keywords. An invention of displaying side by side is disclosed.

Patent Document 2 discloses an invention in which a key composed of text data is extracted from media data such as images and music, and a database in which the media data, the key, a site on a network, and the like are associated is disclosed.
Japanese Patent Application Laid-Open No. 2000-231569 JP 2000-76300 A

  Methods for performing search using text data such as the inventions described in Patent Documents 1 and 2 are widely known in the technical field of Internet search. In addition, as in image search, images on the Internet are often searched from text information.

  However, the above-described conventional technique has a problem that it is impossible to check on which page on the Internet the image already on hand is placed. Therefore, a method for efficiently examining an illegal reprint of an image owned by the copyright owner who owns the copyright of the image has not yet been provided.

  The present invention has been made in view of such circumstances, and an object thereof is to provide an image search system, a crawling device, and an image search device that can easily know where a desired image is posted on the Internet. And

  The image search system according to claim 1 is an image search system including a crawling device and an image search device, and the crawling device is included in a URL of a web page from the Internet and the web page. Calculated by the first identification information calculating means, the first identification information calculating means for calculating the identification information of the image based on the image data of the image collected by the crawling means, and the first identification information calculating means. A database that stores the identified identification information in association with the URL of the web page on which the image serving as the basis of the identification information was posted, and the URL of the web page collected by the crawling means; The image search device includes search image acquisition means for acquiring an image to be searched, and search image acquisition means. A second identification information calculation unit that calculates identification information of the image from the acquired image, and whether or not the same identification information as the identification information calculated by the second identification information calculation unit is included in the database. When the database contains the same identification information as the identification information calculated by the second identification information calculation means, the URL of the web page associated with the identification information is acquired from the database. Search means; and output means for outputting the URL acquired by the search means.

  According to the image search system of claim 1, the URL of the web page and the image included in the web page are collected from the Internet, and the identification information of the image is calculated from the collected image. The identification information and the URL of the web page on which the image that is the basis of the identification information is associated are stored in the database. Then, an image to be searched is acquired, identification information of the image is calculated from the acquired image, whether or not the same identification information as the calculated identification information is included in the database, and the calculated identification information If the same identification information is included in the database, the URL of the web page associated with this identification information is acquired from the database. As a result, the web page on which the image is posted can be searched using the image itself as a search query. Therefore, it is possible to know the publisher of the image at hand afterwards.

  The image search system according to claim 2 is the image search system according to claim 1, wherein the crawling unit obtains a URL of the image together with the image included in the web page, and the database includes the crawling unit. The URL of the image obtained by the above is stored in association with the identification information of the image, and the search means acquires the URL of the web page and the URL of the image associated with the identification information from the database. .

  According to the image search system of the second aspect, when the URL of the image is stored in the database in association with the identification information, and the same identification information as the identification information of the image to be searched is included in the database, The URL of the web page and the URL of the image associated with this identification information are acquired from the database. As a result, it is possible to know the use and reprint status of the searched image on the Internet.

  The image search system according to claim 3 is the image search system according to claim 1 or 2, wherein the crawling device is configured to determine whether a digital watermark is inserted into the image collected by the crawling means, and whether the digital watermark is In the case where it is inserted, it comprises a digital watermark detection means for detecting information related to the digital watermark, the image search device comprises a digital watermark information acquisition means for acquiring information related to a digital watermark to be searched, and the database Storing the information related to the digital watermark detected by the digital watermark detection means in association with the identification information of the image from which the information related to the digital watermark is detected, and the search means is the electronic watermark acquired by the digital watermark information acquisition means Search whether the database contains the same information as watermark information. When the same information as the information related to the digital watermark acquired by the digital watermark information acquisition means is included in the database, the URL and identification information of the web page associated with the information related to the digital watermark are acquired from the database. It is characterized by doing.

  According to the image search system of the third aspect, information relating to the electronic watermark detected from the image in association with the identification information of the image is stored in the database. If the information related to the digital watermark to be searched is acquired and the same information as the information related to the acquired digital watermark is included in the database, the URL and identification information of the web page associated with the information related to the digital watermark are obtained. Get from database. As a result, even when processing such as overwriting characters on the image or trimming the image is performed, the web page on which the image is posted can be searched. Therefore, it is possible to know the publisher of the image at hand afterwards.

  The image search system according to claim 4 is the image search system according to any one of claims 1 to 3, wherein the identification information is a hash value calculated by applying a hash function to the image. It is characterized by being. Thereby, a web page on which the same image is posted can be reliably searched.

  An image search system according to a fifth aspect is the image search system according to any one of the first to third aspects, wherein the identification information is a feature amount indicating a characteristic unique to the image. As a result, not only the same image but also data such as enlargement / reduction of the image, change of the storage format (for example, jpeg format to bmp format), and the like are different, but the distinction is seen from the user. You can also search for web pages with images that are not connected.

  The image search system according to claim 6 is an image search system including a crawling device and an image search device, and the crawling device is included in a URL of a web page from the Internet and the web page. Crawling means for collecting still images, whether or not a digital watermark is inserted in the still images collected by the crawling means, and a digital watermark for detecting information related to the digital watermark when a digital watermark is inserted And a database that stores the URL of the web page collected by the crawling unit and the information related to the digital watermark detected by the digital watermark detection unit in association with each other. Digital watermark information acquisition hand that acquires information about digital watermark And whether the same information as the digital watermark information acquired by the digital watermark information acquisition means is included in the database, and the same information as the digital watermark information acquired by the digital watermark information acquisition means is A search unit that acquires the URL of the web page associated with the information related to the digital watermark, and an output unit that outputs the URL of the web page acquired by the search unit, if included in the database; It is characterized by that.

  According to the image search system of claim 6, the URL of the web page and the still image included in the web page are collected from the Internet, and whether or not a digital watermark is inserted into the collected still image, If a digital watermark is inserted, information on the digital watermark is detected, and the detected information on the digital watermark and the URL of the collected web page are associated with each other and stored in a database. Then, the information about the digital watermark to be searched is acquired, whether or not the same information as the information about the acquired digital watermark is included in the database, and the same information as the information about the acquired digital watermark is included in the database. If so, the URL of the web page associated with the information related to the digital watermark is acquired. As a result, even when processing such as overwriting characters on the image or trimming the image is performed, the web page on which the image is posted can be searched. Therefore, it is possible to know the publisher of the image at hand afterwards.

  The image search system according to claim 7 is the image search system according to claim 6, wherein the crawling unit acquires a URL of the image together with the image included in the web page, and the database includes the crawling unit. The URL of the image obtained by the above is stored in association with information relating to the digital watermark of the image, and the search means obtains the URL of the web page and the URL of the image associated with the information relating to the digital watermark of the image from the database. It is characterized by doing.

  According to the image search system of the seventh aspect, the URL of the image is stored in the database in association with the information related to the digital watermark detected from the image. When the information related to the digital watermark is acquired as a detection target and the same identification information as the information related to the digital watermark is included in the database, the URL of the web page and the URL of the image associated with the information related to the digital watermark are stored in the database. Get from. As a result, it is possible to know the use and reprint status of the searched image on the Internet.

  A crawling device according to an eighth aspect constitutes an image search system according to any one of the first to seventh aspects.

  According to a ninth aspect of the present invention, there is provided an image retrieval apparatus according to any one of the first to seventh aspects.

  According to the present invention, it is possible to easily know where a desired image is posted on the Internet.

<First Embodiment>
FIG. 1 is a schematic diagram of the overall structure of an image search system 1 according to the first embodiment. The image search system 1 mainly includes a crawling unit 10, a search unit 20, and an image DB (database) 30.

  The crawling unit 10 mainly includes a crawling engine 11 and an image hash calculation device 12. The search unit 20 mainly includes an image input device 21, an image hash calculation device 22, a search device 23, and a search result display. And device 24. The crawling engine 11 is connected to the Internet 40. In addition, the image DB 30 is connected to the crawling engine 11, and the image DB 30 is connected to the search device 23.

  First, the crawling unit 10 will be described.

  The crawling engine 11 connects to a web page on the Internet 40 and acquires the URL of the web page. The crawling engine 11 traces link information described in HTML (Hyper Text Markup Language) or the like, and acquires the URL of the image posted on the web page and its image data.

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

  The crawling engine 11 stores the hash value of the image in the image DB (database) 30 in association with the URL of the web page from which the image is acquired, the URL of the image, and the like. The crawling engine 11 automatically circulates the Internet 40 and recursively performs this process, thereby continuously increasing the image DB 30.

  The data structure of the image DB 30 is shown in FIG. The image DB 30 stores the URL of the web page on which the image is posted, the URL of the image posted on the web page, and the hash value of the image in association with each other.

  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 “12345678” is stored in the first row, the fourth row, and the seventh row, indicating that these three are the same image.

  Next, the search unit 20 will be described.

  The image input device 21 acquires an image to be searched and outputs it to the image hash calculation device 22.

  The image hash calculation device 22 receives the image output from the image input device 21 and calculates a hash value of the image using a hash function such as MD5 or SHA-1.

  The search device 23 searches whether or not the image DB 30 includes a hash value that matches the hash value that is the same as the hash value output from the image hash calculation device 22, and the URL of the web page associated with the hash value and The URL of the image is acquired from the image DB 30.

  For example, it is assumed that the hash value “123456789” is calculated by the image hash calculation device 22 from the image acquired by the image input device 21 and the hash value “123456789” is input from the image hash calculation device 22 to the search device 23. As illustrated in FIG. 2, since the hash value “123456789” is stored in the first row, the fourth row, and the seventh row in the image DB 30, the search device 23 performs the first row, the fourth row, and the seventh row. The URL of the web page and the URL of the image stored in the line are acquired from the image DB 30.

  The search device 23 outputs the search results to the search result display device 24 in a list format. The search result display device 24 is a liquid crystal display capable of color display, and the search result output from the search device 23 is displayed on the search result display device 24.

  An example of the search result displayed on the search result display device 24 is shown in FIG. In the search result display device 24, the URL “http: //foo/index.html” of the web page stored in the first line of the image DB 30 and the URL “http: // foo / img. jpeg ”, web page URL“ http: //bar/index.html ”stored in the fourth line, and image URL“ http: //foo/img.jpeg ”, stored in the seventh line. The URL of the web page “http: //hoge/img.jpeg” and the URL of the image “http: //hoge/img2.jpeg” are displayed as a list.

  Thereby, it turns out that the image acquired with the image input device 21 is published on three web pages. In addition, images posted on the web pages “http: //foo/img.jpeg” and “http: //bar/index.html” are linked to the same image URL “http: //foo/img.jpeg”. Although it is a link, it can be seen that the image posted on the web page “http: //hoge/img.jpeg” is copied and reprinted from the image data itself.

  According to the present embodiment, it is possible to search for a web page on which an image is posted by using the image itself instead of a keyword as a search query. Therefore, it is possible to know the publisher of the image at hand afterwards. In addition, it is possible to know the use and reprint status of the image on the Internet.

  Further, according to the present embodiment, since the image itself is used as a search query, it is possible to prevent detection of a web page on which a different image is posted. In addition, since the user does not need to input a keyword or the like, usability can be improved.

  In the present embodiment, the URL of the web page and the URL of the image are displayed on the search result display device 24 as the search results. However, the display of the URL of the image is not essential, and only the URL of the web page is retrieved. You may make it display on the display apparatus 24. FIG. In this case, the URL of the web page on which the image is posted and the hash value of the image may be stored in the image DB 30 in association with each other.

  Note that this embodiment can be applied not only to still images but also to moving images, and also to various contents other than images such as music.

<Second Embodiment>
In the first embodiment, the web page on which the image is posted is searched based on the hash value. However, the method for searching the web page on which the image is posted using the image itself as a search query is not limited to this. Absent.

  The second embodiment is a form in which a web page on which an image is posted is searched based on a feature amount indicating a characteristic unique to the image such as an image pattern, color, and brightness. Hereinafter, the image search system 2 according to the second embodiment will be described. Hereinafter, the same parts as those in the first embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

  As shown in FIG. 4, the image search system 2 mainly includes a crawling unit 10-1, a search unit 20-1, and an image DB 31. The crawling unit 10-1 mainly includes a crawling engine 13 and an image feature amount calculation device 14, and the search unit 20-1 mainly includes an image input device 21, an image feature amount calculation device 25, and a similarity search. A device 26 and a search result display device 24 are included. The crawling engine 11 is connected to the Internet 40. An image DB 31 is connected to the crawling engine 11, and the image DB 31 is connected to the similarity search device 26.

  First, the crawling unit 10-1 will be described.

  The crawling engine 13 connects to an electronic document on the Internet 40 and acquires the URL of the web page. The crawling engine 13 traces link information described in HTML or the like, and acquires the URL of the image posted on the web page and its image data. The images collected by the crawling engine 13 are output from the crawling engine 13 to the image feature amount calculation device 14.

  The image feature amount calculation device 14 calculates the feature amount of the image input from the crawling engine 13. 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, as shown in FIG. 5, the image feature quantity calculating device 14 includes an upper left area (area 1), a lower left area (area 2), an upper right area (area 3), and The average value of the colors (for example, Y component) of the five regions of the lower right region (region 4) and the central region (region 5) of the image is calculated, and region 1, region 2, region 3, region A five-dimensional vector using the average values of the colors of 4 and the region 5 is calculated as a feature amount.

[Equation 1]
Feature amount = (average value of region 1, average value of region 2, average value of region 3, average value of region 4, average value of region 5)
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 feature amount calculated by the image feature amount calculation device 14 is output to the crawling engine 13. The crawling engine 13 stores the feature amount of the image in the image DB 31 in association with the URL of the web page from which the image is acquired, the URL of the image, and the like. The crawling engine 13 automatically circulates the Internet 40 and recursively performs this process, thereby continuously increasing the image DB 31.

  The data structure of the image DB 31 is shown in FIG. The image DB 31 stores the URL of the web page on which the image is posted, the URL of the image posted on the web page, and the feature amount of the image in association with each other.

  Next, the search unit 20-1 will be described.

  The image feature amount calculation device 25 calculates a feature amount from the image output from the image input device 21 by the same method as the image feature amount calculation device 14.

  The similarity search device 26 searches whether the image DB 31 includes a feature amount that is the same or substantially the same as the feature amount output from the image feature amount calculation device 25, and the URL of the web page associated with the feature amount. The URL of the image is acquired from the image DB 31.

  For example, it is assumed that the feature value “12345” is calculated in the image feature value calculation device 25 from the image acquired by the image input device 21, and the feature value “12345” is input from the image feature value calculation device 25 to the similarity search device 26. . As shown in FIG. 6, since the feature amount “12345” is stored in the first row, the fourth row, and the seventh row in the image DB 31, the similarity search device 26 uses the first row, the fourth row, and the The URL of the web page and the URL of the image stored in the seventh line are acquired from the image DB 31.

  At this time, a slight error of about ± 1 digit is allowed, and the URL of the web page and the URL of the image associated with the same and substantially the same feature amount are acquired. This ensures acquisition of the URL of the web page and the URL of the image associated with a similar image that has been processed for enlargement / reduction, but is indistinguishable from the original image from the user's perspective. Can do.

  The similarity search device 26 outputs the search results to the search result display device 24 in a list format. The search result display device 24 is a liquid crystal display capable of color display, and the search result output from the similarity search device 26 is displayed on the search result display device 24.

  According to the present embodiment, it is possible to search for a web page on which an image is posted by using the image itself instead of a keyword as a search query. Therefore, it is possible to know the publisher of the image at hand afterwards. In addition, it is possible to know the use and reprint status of the image on the Internet.

  Further, according to the present embodiment, since the image itself is used as a search query, it is possible to prevent detection of a web page on which a different image is posted. In addition, since the user does not need to input a keyword or the like, usability can be improved.

  In addition, according to the present embodiment, because the search is performed based on the feature amount indicating the feature of the image itself, processing such as enlargement / reduction of the image, change of the storage format (for example, jpeg format to bmp format), etc. Even if it is performed, it is possible to surely search for a web page on which an image different from data is posted although it cannot be distinguished from the user.

  In the present embodiment, a multidimensional vector is used as the feature amount. However, the present invention is not limited to this, and various types composed of about 5 to 100 characters can be used.

  In this embodiment, the URL of the web page and the URL of the image are output as the search results, but only the URL of the web page may be output.

  Note that this embodiment can be applied not only to still images but also to key frames of moving images.

<Third Embodiment>
In the first embodiment and the second embodiment, a web page on which an image is posted is searched based on identification information calculated from image data such as a hash value and a feature amount. These methods are simple and reliable methods, but cannot cope with processing such as overwriting characters on an image.

  In the third embodiment, a web page on which an image is posted is searched based on a digital watermark embedded in the image. The image search system 3 according to the third embodiment will be described below. Hereinafter, the same parts as those in the first embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

  As shown in FIG. 7, the image search system 3 mainly includes a crawling unit 10-2, a search unit 20-2, and an image DB 32. The crawling unit 10-2 mainly includes a crawling engine 15 and a watermark detection device 16. The search unit 20-2 mainly includes a watermark information input device 27, a search device 28, and a search result display device 24. Consists of. The Internet 40 is connected to the crawling engine 15. An image DB 32 is connected to the crawling engine 15, and the image DB 32 is connected to the search device 28.

  First, the crawling unit 10-2 will be described.

  The crawling engine 15 connects to an electronic document on the Internet 40 and acquires the URL of the web page. The crawling engine 15 traces link information described in HTML or the like, and acquires the URL of the image posted on the web page and its image data. The images collected by the crawling engine 15 are output from the crawling engine 15 to the watermark detection device 16.

  The watermark detection device 16 detects whether a digital watermark is embedded in the image input from the crawling engine 15 or what kind of digital watermark is embedded when the digital watermark is embedded. Various methods for detecting a digital watermark are known and will not be described. When a digital watermark is detected by the watermark detection device 16, a character string of the detected digital watermark is output to the crawling engine 15.

  When the watermark detection device 16 detects a digital watermark, the crawling engine 15 associates the detected digital watermark character string with the URL of the web page from which the image was acquired, the URL of the image, and the like. When the digital watermark is not detected by the watermark detection device 16 stored in the DB 32, the URL of the web page from which the image is acquired is associated with the URL of the image and stored in the image DB 32. The crawling engine 15 automatically circulates the Internet 40 and recursively performs this process, thereby continuously increasing the image DB 32.

  The data structure of the image DB 32 is shown in FIG. The image DB 32 stores the URL of the web page on which the image is posted, the URL of the image posted on the web page, and the character string of the digital watermark embedded in the image in association with each other. In the case shown in FIG. 8, the first line, the second line, the fourth line, the sixth line, and the seventh line store the character string “Copyright fujifilm”. This indicates that a digital watermark composed of the character string “Copyright fujifilm” has been detected. The third and fifth lines indicate that no digital watermark character string is stored, and no digital watermark was detected from these two images.

  Next, the search unit 20-2 will be described.

  The watermark information input device 27 inputs information about a digital watermark embedded in an image to be detected, for example, a character string of the digital watermark.

  The search device 28 searches the image DB 32 for whether or not the character string identical to the character string output from the watermark information input device 27 is included, and the URL of the web page associated with the character string, the URL of the image, Is acquired from the image DB 32.

  For example, it is assumed that a character string “Copyright fujifilm” is input from the watermark information input device 27 to the search device 28. As shown in FIG. 8, the character string “Copyright fujifilm” is stored in the first row, the second row, the fourth row, the sixth row, and the seventh row in the image DB 32. The URL of the web page and the URL of the image stored in the first line, the second line, the fourth line, the sixth line, and the seventh line are acquired.

  The search device 28 outputs the search results to the search result display device 24 in a list format. The search result display device 24 is a liquid crystal display capable of color display, and the search result output from the search device 28 is displayed on the search result display device 24.

  According to this embodiment, even when processing such as overwriting characters on an image or trimming an image is performed by using a digital watermark, a web page on which the image is posted is searched. can do. Therefore, it is possible to know the publisher of the image at hand afterwards. In addition, it is possible to know the use and reprint status of the image on the Internet.

  In the present embodiment, the image in which the digital watermark is not detected is also stored in the image DB. However, only the image in which the digital watermark is detected may be stored in the image DB.

  Note that this embodiment can perform detection with higher accuracy when used in combination with the first embodiment or the second embodiment.

  Note that the present invention may be provided as a device that includes all of the crawling unit, the search unit, and the image DB, or may be provided as a system that includes 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.

1 is a schematic diagram of an image search system 1 to which the present invention is applied. It is an example of the data preserve | saved at image DB30 of the image search system 1. FIG. It is a display example of the search result of the image search system 1. 1 is a schematic diagram of an image search system 2 to which the present invention is applied. It is a figure explaining calculation of a feature-value. It is an example of the data preserve | saved at image DB31 of the image search system 2. FIG. 1 is a schematic diagram of an image search system 3 to which the present invention is applied. It is an example of the data preserve | saved at image DB32 of the image search system 3. FIG.

Explanation of symbols

  1, 2, 3: Image search system 10, 10-1, 10-2: Crawling unit, 11, 13, 15: Crawling engine, 12: Image hash calculation device, 14: Image feature amount calculation device, 15: Input Unit: 16: watermark detection device, 20, 20-1, 20-2: search unit, 21: image input device, 22: image hash calculation device, 23, 28: search device, 24: search result display device, 25: Image feature amount calculation device, 26: similarity search device, 27: watermark information input device, 30, 31, 32: image DB, 40: Internet

Claims (9)

An image search system composed of a crawling device and an image search device,
The crawling device is
A crawling means for collecting a URL of a web page and an image included in the web page from the Internet;
First identification information calculation means for calculating identification information of the image based on image data of the image collected by the crawling means;
The identification information calculated by the first identification information calculation means, and the URL of the web page on which the image serving as the basis of the identification information was posted, and the URL of the web page collected by the crawling means A database to be stored in association,
The image search device includes:
Search image acquisition means for acquiring an image to be searched;
Second identification information calculation means for calculating identification information of the image from the image acquired by the search image acquisition means;
It is searched whether the same identification information as the identification information calculated by the second identification information calculation means is included in the database, and the same identification information as the identification information calculated by the second identification information calculation means is If included in the database, search means for obtaining the URL of the web page associated with the identification information from the database;
Output means for outputting the URL acquired by the search means;
An image search system comprising: The crawling means acquires the URL of the image together with the image included in the web page,
The database stores the URL of the image acquired by the crawling means in association with the identification information of the image,
The image search system according to claim 1, wherein the search unit acquires the URL of a web page and the URL of an image associated with the identification information from the database. The crawling device includes a digital watermark detection unit that detects whether or not a digital watermark is inserted in the image collected by the crawling unit, and detects information related to the digital watermark when the digital watermark is inserted,
The image search device includes digital watermark information acquisition means for acquiring information related to a digital watermark to be searched,
The database stores information related to the digital watermark detected by the digital watermark detection means in association with identification information of an image in which the information related to the digital watermark is detected,
The search means searches for whether the same information as the information related to the digital watermark acquired by the digital watermark information acquisition means is included in the database, and is the same as the information related to the digital watermark acquired by the digital watermark information acquisition means 3. The image search according to claim 1, wherein when the information is included in the database, the URL and identification information of the web page associated with the information related to the digital watermark are acquired from the database. system.   The image search system according to claim 1, wherein the identification information is a hash value calculated by applying a hash function to the image.   The image search system according to any one of claims 1 to 3, wherein the identification information is a feature amount indicating a feature unique to the image. An image search system composed of a crawling device and an image search device,
The crawling device is
Crawling means for collecting a URL of a web page from the Internet and a still image included in the web page;
Whether or not a digital watermark is inserted in the still image collected by the crawling means, and if a digital watermark is inserted, a digital watermark detection means for detecting information relating to the digital watermark;
A database that stores URLs of web pages collected by the crawling means in association with information about the digital watermark detected by the digital watermark detection means, and
The image search device includes:
Digital watermark information acquisition means for acquiring information related to a digital watermark to be searched;
It is searched whether or not the same information as the digital watermark information acquired by the digital watermark information acquisition means is included in the database, and the same information as the digital watermark information acquired by the digital watermark information acquisition means is stored in the database. If included, search means for acquiring the URL of the web page associated with the information related to the digital watermark;
Output means for outputting the URL of the web page acquired by the search means;
An image search system comprising: The crawling means acquires the URL of the image together with the image included in the web page,
The database stores the URL of the image acquired by the crawling means in association with information on the digital watermark of the image,
The image search system according to claim 6, wherein the search unit acquires the URL of the web page and the URL of the image associated with the information related to the digital watermark of the image from the database.   A crawling device constituting the image search system according to claim 1.   An image search apparatus constituting the image search system according to claim 1.

Images