JP2006106404A - Advertisement display method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To realize highly effective advertisement display which reflects tastes of individual image owners, which are obtained from a plurality of images. <P>SOLUTION: A common element of a plurality of images is extracted, and an advertisement corresponding to the common element is selected from an advertisement database and is displayed. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an advertisement display method.

With the widespread use of digital cameras, camera-equipped mobile phones, and large-capacity, inexpensive recording media, a large number of image browsing and searching are becoming common. In addition, in order to increase the effectiveness of advertisements, methods for associating images with advertisements are being proposed. There have been proposed a method of adding an advertisement text to an image and a method of performing image processing to extract an advertisement feature amount. For example, in Patent Document 1, information for specifying a person such as gender, age, and clothes is extracted from a person image and used for displaying an advertisement. By such means, it is possible to effectively advertise according to individual persons.
JP 2002-73321 A

  However, the method as described in Patent Document 1 focuses on only information obtained from a single image, and cannot handle information captured from a plurality of images. In recent years, a situation where one user possesses and browses a large number of images is common. Therefore, if attention is paid to common features in a plurality of images possessed by a user, individual user preferences that cannot be obtained from a single image can be obtained. By using this information, it is possible to display more effective advertisements.

  Therefore, this proposal aims to realize a more effective display of advertisements by using common features in a plurality of images.

  The notice display method of the present invention is characterized by using the following components.

  Corresponds to the image group recorded in the recording medium possessed by the image owner, image search means for comparing images by similarity image search, judgment of coincidence, representative image group used for similar image search, and common elements of multiple images Database of advertisements.

  It is possible to select whether or not to display the advertisement.

  The display number of the advertisement can be selected.

  A group of images recorded in a recording medium shared by a plurality of image holders on a server in the network or the like is a target for the similar image search.

  It is possible to add a common element extraction method corresponding to the advertisement display as needed.

  The image owner can use the recording area on the network managed by the advertiser by allowing the image owner to display the advertisement.

  The larger the quantity of advertisement display permitted by the image owner, the more the image owner can use the recording area on the network.

  The common element of the similar image search result is quantitatively evaluated, and the advertisement is displayed larger as the degree of correspondence with the corresponding advertisement in the advertisement database is higher.

  First, the present realization means according to claim 1 has an effect of grasping a user's preference that cannot be obtained from a single image and displaying an advertisement that arouses the user's interest. Moreover, when the content of the possessed image changes, it is possible to recognize the change as a change in preference for each user and reflect it as a change in the displayed advertisement. Moreover, since it becomes possible for an advertiser to display an advertisement to a user who is more interested in the contents of the advertisement, the advertisement efficiency is increased. In addition, since the image group as a result of the search by the user is considered to strongly reflect the user's interest at the time of search, there is an effect that more effective advertisement display is possible.

  Secondly, the realization means according to claim 2 has an effect of suppressing the display when the user does not want to display the advertisement.

  Thirdly, by the realization means according to claim 3, and fourthly, by the realization means according to claim 4, by displaying an advertisement from the common element in the image in the recording medium on the network that is widely disclosed, It has an effect that an effective advertisement can be displayed to an image viewer other than the image owner who has a common preference with the image owner.

  Fifth, the realization means according to claim 5 enables the advertiser to add or change the handling of the common elements of the similar image search results for effective advertisement as needed.

  Sixth, the realization means according to claim 6 has an effect that the advertiser can display the advertisement and the user can use the recording area on the network at a low cost.

  Seventh, the realization means according to claim 7 enables the advertiser to increase the possibility of displaying more advertisements, and the user can select the amount of recording area on the network by selecting the amount of advertisement display. It has the effect.

  Eighth, the realization means according to claim 8 has an effect that the advertiser can more emphasize the advertisement to the user who is considered to be highly effective. In addition, the advertisement can be emphasized more than other advertisements displayed at the same time.

  The best mode in the present invention is as follows.

(First embodiment)
In the present embodiment, the image file group is assumed to exist in a general HDD in the PC. Recent HDDs usually have a large capacity of several tens of gigabytes, and the number of stored image files has become enormous with the spread of digital cameras and camera-equipped mobile phones.

  First, the flow from similar image search to advertisement display in the present embodiment will be described with reference to FIG. Reference numeral 10 denotes an image file group made up of a large number of image files stored in the HDD, and reference numeral 11 denotes an image group obtained as a result of a similar image search. Although a search method for performing similar image search is not limited, a general similar image search method uses a histogram that expresses the contrast of the entire image or contour extraction that focuses on the frequency components of the image. In addition, the use of color information and information such as the shape and inclination of the subject are also used for similarity search. In order to improve the search effect, it is necessary to use a search method that uses more advanced features such as person recognition and facial expression recognition.

  Next, suppose that the common element detected by the detection method which detects the common element relevant to the advertisement which an advertiser wants to display is extracted as 12.

  Reference numeral 12 denotes a common element obtained from the similar image search result image group. In this embodiment, it is assumed that common elements having the contents of “animal” and “pet” are obtained. More specifically, the advertiser prepares an algorithm capable of extracting elements such as “animal” and “pet” from the image, and in this example, the similar image search result 11 is an image group including many of these elements.

  Since the act of performing a similar image search is an act of selecting a group of images that the user is interested in, it is considered effective to select the advertisement by extracting the common element 12 from the group of images 11 of the similar image search result. .

  The numerical value in the common element 12 in FIG. 1 indicates the number of common elements. Here, the total amount of the “animal” elements included in the image group 11 obtained as a similar image is “50”, and the “pet” element. Is represented as “20”.

  Advertisers devise feature extraction methods corresponding to the contents of advertisements and extract them from the image group, but feature extraction is not used for image retrieval, so it is not necessary to perform in real time, it is performed in the background as needed do it. Also, consider a case where an advertiser wants to place a more limited advertisement. For example, in this embodiment, the common element “pet” is generally narrower than the “animal”, and the “pet” is designed to devise a common element corresponding to an advertisement for only a specific type of animal. It is considered that more detailed feature extraction is required. As detailed feature extraction takes more time to process an image, it is not necessary to perform it in real time, so the advertiser does not need to consider the feature extraction load relatively.

  Next, a flow for selecting an advertisement will be described with reference to FIG. An advertisement related to the common element 12 “animal” and “pet” is selected from the advertisement database 13. The correspondence between the common element 12 and the advertisement in the advertisement database 13 is one-to-one. For example, in this embodiment, since the advertiser of the advertisement A (zoo) prepares an algorithm for extracting the “animal” element and extracts the “animal” element by the algorithm, the “animal” element and the advertisement A (zoo) ) Has been made. In addition, it is assumed that the advertisement D (pet shop) corresponds to the “pet” element. In this embodiment, the advertisement A and the advertisement D corresponding to “animal” and “pet” having a high value as the common element 12 are selected and displayed as advertisements.

  The advertisement in the advertisement database can be changed by the advertiser at any time, and the contents of the advertisement can be freely changed or deleted if the contents of the advertisement have a time limit.

  Next, FIG. 3 shows an example of actual advertisement display. Reference numeral 30 denotes a display medium, which is a display of a PC in this embodiment. Reference numeral 31 denotes an image group obtained as a result of performing a similar image search from an image group possessed by the user. The image group 10 in FIG. 1 is displayed on the display medium 30. Reference numeral 32 denotes an advertisement corresponding to “animal” and “pet”, which are common elements of the image group 31, and the advertisement is assumed to be a banner advertisement here.

  In this way, the elements that are commonly seen in multiple images are thought to reflect the user's preferences, so selecting and displaying advertisements corresponding to the common elements will result in more effective advertisements. It is done.

  In this embodiment, the common element is extracted from the image of the similar image search result, but the common element may be extracted from all the images possessed by the user instead of the result of the similar image search, and the advertisement may be displayed.

  In the present embodiment, the image file group is stored in the HDD. However, in recent years, a flash memory used for a digital camera or a camera-equipped mobile phone, or a DVD that is generally written by a PC or a DVD recorder is generally used. The medium is also a large-capacity recording medium, and can be considered as a recording medium that stores many images. When a plurality of users share an image on a recording medium on the network, it is possible to display an effective advertisement for a plurality of users having a common preference.

  Furthermore, although not in the present embodiment, for example, “can color” and “product name logo” for beverages, and “shape”, “color”, “brand logo”, etc. for clothing are common. Listed as an element. "Product name logo" and "brand logo" can be extracted by character recognition, but such character recognition is also considered as a feature extraction method.

  In addition, it is conceivable that information added as metadata to each image is used as a criterion for advertisement display together with common elements extracted by image processing. As metadata added to images, various information such as shooting date and time, shooting location, number of times of browsing, image size, shooting conditions, etc. can be considered, but advertisers can freely combine these information to determine effective advertising display criteria Can be thought of as

  When there is a margin in the image display area, the advertisement effect can be further enhanced by increasing the advertisement to be displayed as the common element value increases. As an example, in the case of common elements such as “animal” and “pet” in the above embodiment, since “animal” is more common when the amount of common elements is compared, the advertisement to be displayed is enlarged.

(Second embodiment)
This embodiment is not limited to a PC, but can be implemented with a digital camera, a mobile phone, a DVD recorder, and a television connected to the DVD recorder. Reference numeral 40 in FIG. 4 denotes a digital camera, and the image display medium is a liquid crystal display on the back side of the digital camera. A liquid crystal display is generally used as a display medium in digital cameras and mobile phones, but there is less image information that can be displayed compared to the display area of a PC or television. For this reason, since it is common that a region used for displaying an advertisement cannot be secured widely, in the present embodiment in FIG. 4, a small banner is displayed below an image including many common elements.

It is a schematic diagram from similar image search to advertisement display in a first embodiment. It is a schematic diagram until an advertisement is selected from the common elements in the first embodiment. It is a display example in the display medium in 1st embodiment. It is an example of a display in the display medium in 2nd embodiment.

Explanation of symbols

10 Image possessed by user 11 Image group of similar image search result
12 Common elements
13 Advertising database
30 Image display media
31 display image group 32 banner advertisement
40 Digital camera

Claims (8)

A group of images recorded in a recording medium possessed by the image owner;
Image search means for comparing images by similarity image search and determining matching;
Representative image group used for similar image search;
An ad database consisting of ads corresponding to the common elements of multiple images,
Ad display method using.   The advertisement display method according to claim 1, wherein whether or not to display the advertisement can be selected.   The advertisement display method according to claim 1, wherein a display number of the advertisement can be selected.   2. The advertisement display method according to claim 1, wherein a group of images recorded in a recording medium shared by a plurality of image holders on a server in a network or the like is a target of the similar image search.   The advertisement display method according to claim 1, wherein a common element extraction method corresponding to the advertisement display can be added at any time.   The advertisement display method according to claim 1, wherein the image owner can use the recording area on the network managed by the advertiser by allowing the image owner to display the advertisement.   2. The advertisement display method according to claim 1, wherein the larger the quantity of advertisement display permitted by the image owner, the more the image owner can use the recording area on the network.   The advertisement display method according to claim 1, wherein the common element of the similar image search result is quantitatively evaluated, and the advertisement is displayed larger as the degree of correspondence with the corresponding advertisement in the advertisement database is higher.

Images