JP2002175311A - Method and device for registering video information, and method and device for retrieving the video information - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide technology capable of registering video information for realizing high retrieving accuracy in a video database. SOLUTION: A video featured value is extracted from an image to be registered, the extracted video featured value is embedded in the image to be registered as a format, corresponding to the identification(ID) information of an extraction algorithm and the image in which the video featured value is padded is registered in the video data base by a format corresponding to the video featured value. The video featured value, embedded in a reading image, is detected, an image, in which a video featured value similar to the detected video featured value is embedded, is retrieved from the video data base and outputted. Thus high retrieval accuracy can be realized in accordance with the constitution, by solving the problem of conventional technology which adopts constitution for outputting the video featured value and the image by a communication data format linking both the featured value and the image.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video information registration method and apparatus for registering video information realizing a high search accuracy in a video database, and a video information registered by the video information registration method and apparatus. The present invention relates to a video information search method and apparatus for enabling a search.

[0002]

2. Description of the Related Art A conventional video search will be described with reference to the Internet video search shown in FIG.

[0003] As shown in this figure, the end user side
Equipped with a means for searching for video media using a terminal device, and the database on the network side is TCP
A target content (video) is obtained while viewing browsing information displayed on a homepage (HP) using a communication procedure based on / IP.

(1) Configuration and operation of the HP unit on the network side The HP unit has a receiving function of receiving search information from the end user, and a transmission function for downloading searched or browsed images. A communication interface consisting of functions is provided, and a video browsing function for operating a database provided in a form linked to the communication interface, and a content transfer function between the communication interface and the database for storing video. A video management function for managing search information is provided.

[0005] In this video browsing function, a still image which is a part of a video to be browsed is stored.
Based on the URL information input from the communication interface,
It manages browse images specified by the user and database access information linked to the browse images.

(2) Configuration and Operation of Database On the other hand, the database is provided with a video content storage function for storing content (video), and a video management function for searching and storing management of the video content storage function.

[0007] Search ID information in the video management information (In the present invention, information indicating the characteristics of the content is referred to as a search ID
Information) and the content stored by the video content storage function are linked, and the video management function of the database selects and outputs the content based on the access information from the video management function of the HP unit, and provides the end user with the information. On the other hand, the content is downloaded.

As described above, since the content and the search information as the ID information of the content are associated with each other via the video management function, when the content is transmitted between the HP unit and the database, It is necessary to communicate in a communication data format in which the search ID information and the content are linked.

In the prior art, specifically, using a data format in which a header portion contains search ID information, a search ID is used.
Information and content are linked for communication.

[0010]

However, when a method of communicating in a communication data format in which search ID information and content are linked as in the prior art is used, the construction of a browsing environment or the like is difficult. When the content sent from the database is subjected to format conversion or image editing processing, there is a problem that the search ID information recorded in the header portion of the communication data format may be destroyed.

From this, there is a problem that the video pointed to by the selected browsing video cannot be searched.

Further, when a method of communicating search ID information and content in a communication data format is used as in the prior art, the cost and labor for maintaining and operating the management system are enormous. There is a problem that it becomes something.

The present invention has been made in view of the above circumstances, and provides a new video information registration method and apparatus capable of registering video information realizing high retrieval accuracy in a video database, and providing the video information registration method. It is an object of the present invention to provide a new video information search method and apparatus that enables video information registered by a registration method and apparatus to be searched.

[0014]

In order to achieve this object, a video information registration method according to the present invention uses a plurality of extraction algorithms to extract a video feature from a video to be registered. The video features that have been extracted, while corresponding to the identification information of the extraction algorithm used to extract them,
The process of embedding in the video to be registered using digital watermark,
Registering the video with the embedded video feature in the video content storage unit while associating the video with the embedded video feature.

In order to achieve this object, a video information search method according to the present invention comprises the steps of: inputting a browsing video in which a video feature is embedded using a digital watermark; The process of detecting the video features stored in the video content storage unit, and accessing the video content storage unit described above, the video features stored in the video content storage unit that are similar to the detected video feature It is configured to include a step of searching, and a step of reading and outputting, from the video content storage unit, a video in which the video feature amount associated with the searched video feature amount is embedded, and outputting the video.

According to this configuration, according to the video information registration method of the present invention, the video feature is embedded in the video to be registered in a form corresponding to the identification information of the extraction algorithm, and the video feature is embedded. The obtained video is registered in the video content storage unit in a form corresponding to the video feature amount.

According to this configuration, according to the video information search method of the present invention, a video feature having a video feature similar to the video feature detected from the browsed video is embedded from the video content storage unit. The retrieved video is retrieved and output.

As described above, according to the present invention, the video with the embedded video feature is registered in the video database, and the video with the embedded video feature is retrieved from the video database and output. Therefore, it is possible to solve the problem of the related art that adopts a configuration in which a video feature amount and a video are output in a communication data format in which the video is linked.

[0019]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail according to embodiments.

FIG. 1 shows an embodiment of the overall configuration of a video search system provided with the present invention.

As shown in this figure, a video search system equipped with the present invention comprises a search system 1 located on the network side and a terminal device 2 located on the end user side.
It is composed of

The search system 1 includes an HP (homepage) unit 10 and a database 11, while the terminal device 2 includes an HP access unit 20 and a download correspondence unit 21.

Next, details of these units will be described.

(1) Configuration of the Search System 1 (a) Configuration of the HP Unit 10 The HP unit 10 provides a home page as an interface to the end user using the database 11.

The HP unit 10 includes a GUI (Graphical User Interface) 100 for operating the database 11, a browse video storage unit 101 for storing browse videos to be searched, and a search sent from the user side. A receiving unit 102 for receiving information and a video transmitting unit 103 for downloading a video or a browsing video retrieved from the database 11 are provided.

The browsing video downloaded by the video transmitting unit 103 includes a part of a moving image, that is, a frame image (still image) or a short continuous frame.
Attributes that are features of the video to be searched are embedded with digital watermarks in these browsed videos. The browsed video includes not only a part of the moving image but also a still image such as a poster gravure exhibited when the movie was released.

(B) Configuration of Database 11 A media search having a digital watermarking function, which is a feature of the present invention, is applied to the database 11. In the present embodiment, the database 11 is installed in the search system 1 on the network side.

The database 11 includes a video content storage section 110, an input interface section 111, an output interface section 112, and a utility section 113.
And are provided.

(A) In the video content storage unit 110,
Videos to be searched are stored. In particular,
The original video, search ID information stored in a form linked to the original video (information indicating characteristics of the video extracted from the original video), and the search ID information embedded in the original video with a digital watermark. And the ID-embedded video created by the operation.

Here, the reason why the original video is stored in addition to the ID-embedded video is that, depending on the digital watermark algorithm, only the original video may not be extracted from the ID-embedded video. The reason why the search ID information is also stored in addition to the ID-embedded video is that if the search ID information is extracted from each of the ID-embedded videos, the search performance is degraded.

(B) The input interface unit 111 is an interface for inputting a viewing video to the database 11. The input interface unit 111 detects the digital watermark information embedded in the browse video, which is a non-linguistic search information source, so that the search I
An electronic watermark detection function for generating D information is provided.

(C) The output interface unit 112 is an interface for outputting the video information searched based on the search ID information from the database 11.

(D) The utility unit 113 manages data of the database 11 and controls data input / output.
The utility unit 113 includes a digital watermark embedding function, a digital watermark detection function, and a video feature extraction function that are required when adding / updating new video information to the video content storage unit 110 and performing maintenance on the stored information. ing.

(2) Configuration of Terminal Device 2 The end user uses the terminal device 2 to search for video media.

The terminal device 2 has a TCP / IP
An HP access unit 20 that performs processing such as selection of a browsed video displayed on a homepage and acquisition of a video attribute using
And a downloading support unit 21 for receiving the searched video information transmitted via the network.

Next, a detailed configuration of the database 11 will be described.

FIG. 2 shows an embodiment of the detailed configuration of the database 11. In the figure, those described with reference to FIG. 1 are denoted by the same symbols, and reference numeral 114 denotes a search unit that performs processing for searching for video information stored in the video content storage unit 110.

For the reasons described above, the video content storage unit 110 embeds the original video, the search ID information stored in a form linked to the original video, and the search ID information embedded in the original video with a digital watermark. And the ID-embedded video created as a result.

Here, a link is provided between the ID-embedded video and the original video from which the ID-embedded video was created. Therefore, the link between the ID-embedded video and the search ID information embedded in the video is embedded. Is linked.

In addition to the above-mentioned reasons, it is more efficient to extract the search ID information from the ID-embedded video during the content creation stage and the content search process. Is linked to and stored.

The contents of the search ID information used for searching for the video information stored in the video content storage unit 110 include:
For example, there are the following attributes, all of which are effectively used for search from the viewpoint of improving search accuracy and reducing search speed.

As information representing the physical characteristics (characteristics) of the video, information such as a scene change pattern, a frame change pattern such as an optical flow pattern, and a landscape (including a scenery scene which is more characteristic than others). If there are, there are various things.

It is more advantageous to use, as the pattern information to be added as the search ID information, data obtained by systematizing and classifying the extracted raw data and replacing it with code information or the like, from the viewpoint of reducing the amount of search ID information. Therefore, the present invention employs a configuration in which search ID information having a quantitative value replaced with code information or the like is used.

An example of the search ID information used in the present invention will be described using the video scene shown in FIG. 3 as a specific example.

In FIG. 3, frames #k and # k +
Each of # 1 and # k + 2 is divided into blocks for extracting video features, and the correspondence when comparing between frames is as shown in the figure. In FIG. 3, one frame is divided into eight blocks, but it goes without saying that actually one frame may be divided into a larger number of blocks.

The video scene shown in FIG. 3 is composed of a street tree scenery scene and a beach scenery scene. The street tree scenery scene includes a background part with no motion (or little motion) and a moving object. On the other hand, it is assumed that the beach scene is a night scene, the video signal value composed of YUV is smaller than the street tree scene, and the motion is smaller.

FIG. 4 shows an example in which an image having such two scenes is arranged as information representing physical characteristics (characteristics) of the image.

Here, it is assumed that an algorithm for extracting a feature of an image extracts an inter-frame difference value, a motion vector, an optical flow, and an average value of Y, U, and V signals of pixels. The numerical value in parentheses indicates the number of blocks having a large change in one frame, and the size of the moving object can be grasped from the number of blocks. The comparison target of the inter-frame difference with respect to the frame #k is a frame # k-1 (not shown), and it is assumed that the background of the frame # k-1 is the same as that of the frame #k except for the moving object.

The optical flow is an extraction algorithm for performing a process of performing Hough transform on a motion vector, and has a function of specifying a direction and an angle of a continuously changing motion.

For the scene change portion where the scene is switched from the street tree scene to the beach scene, for example,
It can be specified by determining the inter-frame difference value for all the blocks and counting the number of blocks having a large change. Regarding the motion vector, in the case of a scene change, the frame # k + 1 is compared with the pixel signal level in the frame #k.
May not exist, and at this time, detection is not possible. In addition, the optical flow is interrupted.
The average pixel value is related to the brightness and color of the entire scene, and serves as a criterion for scene change.

Although all of these video feature algorithms are known and can be easily realized, by combining a large number of them, the video features from various viewpoints can be quantified.

Although the video feature shown in FIG. 4 is a process in an image space, a feature in a frequency space may be added. For example, a relationship between the spatial frequency ω and the power spectrum S (ω) is obtained by performing a Fourier transform for each frame. In this relation S (ω) ∝ω- ^D , a positive constant D defining the degree of reduction of S (ω)
It is known that (fractal dimension) contributes to a video landscape (for example, the degree of steepness of a mountain range), and this constant D may be added to the search ID information.

In the present invention, video features are embedded as digital watermarks. In this case, instead of features for each frame, systematically organized and categorized information may be embedded as search ID information. For example, the inter-frame difference value for all frames of the video or the change width and average value of the motion vector, the number of scene changes, statistics of video features such as fractal dimension, and the inter-frame difference value for each video scene Alternatively, a change width of a motion vector, a fractal dimension, or the like may be embedded as search ID information.

As described above, in the case where a plurality of algorithms are used as the algorithm for extracting video features, although the image feature extraction result is a set of many video features, complicated means such as image recognition is necessary. It can be easily realized. If the pattern of the physical characteristic is extreme, the quantified value is an important judgment material to distinguish it from other images.

The feature of the method using such video characteristics is that the search accuracy is surely improved by increasing the evaluation criteria. Although it is difficult to completely describe video identification in the content definition based on human senses such as keywords, a method using such video characteristics is effective in solving the problem.

Of course, it is important to add keywords, comment information, and the like as related to the video contents.
Such information other than the video content itself can also be used as search ID information. Also, for example,
Information irrelevant to video content, such as video production, capacity, distribution, etc., can also be used as search ID information as needed.

By adding such management information (keywords, comments, and other information) not directly related to the video contents to the search ID information, the search accuracy can be further improved.

Returning to the description of FIG.
1 includes an input interface unit 111, a search unit 114, an output interface unit 112, and a utility unit 113 in addition to the video content storage unit 110.

The input interface unit 111 performs a process of detecting search ID information embedded in the browse video using the digital watermark detection unit 1110 when the browse video is selected. From the search ID information stored in the video content storage unit 110,
The output interface unit 112 performs a process of searching for one similar to the searched search ID information, and outputs the ID-embedded video linked to the searched search ID information.

The digital watermark detecting section 1110 of the input interface section 111 has a plurality of watermark detecting functions of different systems. There are various algorithms for embedding digital watermarks that have suitability for the embedding target.
In order to cope with them, watermark detection methods 1 to M having different algorithms are provided.

The digital watermark detection unit 1110 detects a watermark from an input browsing image by a certain algorithm (can be evaluated based on the detected amount of the watermark).
, The watermark is detected by switching to another algorithm. Note that even if the selection of the algorithm is appropriate, the watermark cannot be detected even when the parameter setting of the detection condition is not appropriate. Therefore, parameter selection is controlled together with algorithm selection, and a watermark is detected.

The digital watermark information detected from the browsed video is input as search ID information to the search unit 114 via the I / O unit. If the search unit 114 finds the corresponding ID-embedded video, it is output. The ID embedded video is output via the interface unit 112.

The switching of the detection algorithm in the digital watermark detection unit 1110, the setting of the detection parameters, and the I / O control with the video content storage unit 110 are performed by the database control unit 1 of the utility unit 113 described later.
135.

Next, a detailed configuration of the utility section 113 will be described.

The utility section 113 performs data management of the database 11 and control of data input / output as described above. As an even more important function, it has a function of adding / updating new video information to the video content storage unit 110. At this time, a feature of the video is extracted and embedded in the video using a digital watermark. Then, with the ID embedded video stored in the video content storage unit 110 as a processing target, search ID information (video feature information) embedded in the ID embedded video is detected.
The monitor output is performed.

In order to perform these processes, the utility unit 113 transmits the image feature extracting unit 11 as shown in FIG.
30, an electronic watermark embedding unit 1131, and an I / O unit 1
132, a digital watermark detection unit 1133, a monitor output unit 1134, and a database control unit 1135.

(1) Video Feature Extraction Unit 1130 The video feature extraction unit 1130 performs a process of extracting the feature of a video to be embedded as a digital watermark, and has a plurality (1 to N) of extraction methods with different algorithms. . For example, it has a function of comparing the frames of a video to specify the speed of a scene change or a moving part, a function of extracting a line drawing texture using Hough transform, and the like. A plurality of these algorithms are combined to perform video feature extraction from various evaluation viewpoints.

(2) Digital Watermark Embedding Unit 1131 The digital watermark embedding unit 1131
30 for embedding a video feature output as an electronic watermark into a video. A plurality of (1 to M) watermark embedding methods having different algorithms are provided, and a video feature amount to be embedded and An algorithm suitable for the video at the embedding destination is selected, and embedding processing is performed using the selected algorithm.

When embedding this video feature (search ID information), the identification ID (identification information represented by a number or the like) of the extraction algorithm used to extract the video feature
Is also embedded.

(3) I / O Unit 1132 Utility Unit 113 and Video Content Storage Unit 110
Acts as an interface between and.

(4) Digital Watermark Detection Unit 1133 The digital watermark detection unit 1133
10 by taking in the ID embedded video stored in
This is a process for detecting embedded search ID information (video feature information). When a plurality of watermark detection methods (1 to M) with different algorithms are provided and a certain algorithm cannot detect a watermark, The watermark is detected by switching to another algorithm.

The search ID information detected in this way is fed back to the video content storage unit 110 as needed, and is linked to a video in which search ID information is not embedded or a video in which search ID information is embedded. It is buffered in the form or output to the monitor output unit 1134.

(5) Monitor output unit 1134 This is an interface prepared to monitor the video and search ID information stored in the video content storage unit 110. Further, monitoring of the search ID information detected by the electronic watermark detection unit 1133 and the ID embedded video from which the search ID information is detected are also performed via this interface. The monitor output unit 1134 performs processing for this,
A video format processing function for visualizing the search ID information is provided.

(6) Database control unit 1135 The database control unit 1135
0, digital watermark embedding unit 1131, I / O unit 113
2. Digital watermark detection unit 1133 and monitor output unit 11
34 is performed.

FIG. 5 shows control signals issued by the database control unit 1135. The control contents of the control signal for each functional module are as follows.

(A) Control signal Cont.1 Selection of digital watermark embedding algorithm and setting of embedding parameters (digital watermark key information, watermark embedding strength, etc.).

(B) Control signal Cont.2 Selection of digital watermark detection algorithm and setting of detection parameters (digital watermark key information and the like).

(C) Control signal Cont.3 Utility unit 113 and video content storage unit 110
Interface control of video and search ID information between and.

(D) Control signal Cont.4 Output from video content storage section 110 (video / ID embedded video / search ID information) and digital watermark detection section 11
Switching of output (ID embedded video / search ID information) from 10.

(E) Control signal Cont.5 Selection of video feature extraction algorithm and setting of extraction parameters.

Such control conditions are set for each function module.
Then, the registration of the video information in the video content storage unit 110 and the output of the video information from the video content storage unit 110 are performed.

Finally, an outline of the processing of the present invention described above will be described with reference to FIG.

In the present invention, the video content storage unit 110
When the video information to be registered is given, as shown in FIG. 6, first, as shown in FIG. The video feature amount is extracted from the video.

At this time, when an image is composed of a plurality of frames, processing may be performed so that a systematic arrangement of image features extracted from each frame is extracted as a final image feature.

Subsequently, as shown in the figure, using the digital watermark embedding unit 1131, the extracted video feature amount is identified by the extraction algorithm I used for the extraction.
While taking correspondence with D, it is embedded in the video to be registered using a digital watermark.

For example, a video feature quantity “αα” is extracted by the first extraction algorithm, and a video feature quantity “ββ” is extracted by the second extraction algorithm.
When the video feature amount “γγ” is extracted by the third extraction algorithm, the video feature amount = (ID = 1: α
A video feature having a data structure of α, ID = 2: ββ, ID = 3: γγ) is embedded in a video to be registered using a digital watermark.

In the figure, there is a case where one specified extraction algorithm is operated instead of operating a plurality of algorithms. At this time, it is not necessary to embed the identification ID of the extraction algorithm.

In this digital watermark embedding process,
Management information of the video to be registered (information not directly extracted from the video content, such as keywords and comments,
Other information unrelated to the video content)
In some cases, processing is performed so as to be embedded in a video to be registered using a digital watermark.

In some cases, a plurality of algorithms for embedding a digital watermark are prepared, an optimum one is selected from the algorithms, and a process for embedding a video feature amount in a video to be registered is performed using the optimum algorithm.

Subsequently, as shown in the figure, the video in which the video feature amount is embedded (the video in which the search ID information is embedded) is associated with the embedded video feature amount (the search ID information). And register it in the video content storage unit 110.

As described above, according to the present invention, the video feature is embedded in the video to be registered so as to correspond to the identification information of the extraction algorithm. The content is registered in the video content storage unit 110 in a form corresponding to the amount.

On the other hand, according to the present invention, when a browsing image in which an image feature amount is embedded using an electronic watermark is given, FIG.
First, as shown in the figure, the browsing video is input.

Subsequently, as shown in the figure, a video feature amount (search ID information) embedded in the input browsed video is detected by using the digital watermark detection unit 1110.

Subsequently, as shown in FIG.
By accessing the video content storage unit 110 using No. 14, the video feature amount (search ID information) stored in the video content storage unit 110 is similar to the detected video feature amount (search ID information). Search (usually the most similar one).

For example, the search unit 114 associates the search ID information detected by the digital watermark detection unit 1110 with the identification ID of the extraction algorithm, and calculates (ID = 1: αα, ID = 2: ββ, ID = 3: γγ), while the video content storage unit 1
If the search ID information stored in 10 has a data value of (ID = 1: aa, ID = 2: bb, ID = 3: cc) in association with the identification ID of the extraction algorithm, for example, the distance = [(Αα-aa) ² + (ββ-bb) ² + (γγ
−cc) ² ] The search process is executed by measuring the similarity between the two pieces of search ID information based on the distance defined by ^1/2 .

Subsequently, as shown in the figure, an ID-embedded video associated with the searched video feature amount (search ID information) is read from the video content storage unit 110 and output.

As described above, in the present invention, the video content storage unit 110 searches for a video in which a video feature amount having a video feature amount similar to the video feature amount detected from the browsed video is embedded, and retrieves the video. Is output.

As described above, since the present invention does not use the conventional method of communicating search ID information (video feature amount) and video in a communication data format in which the video is linked, it is necessary to construct a browsing environment. Therefore, even when the format sent from the video database is converted or the image editing process is performed, the disadvantage that the search ID information (video feature amount) is destroyed does not occur.

Although the present invention has been described with reference to the illustrated embodiment, the present invention is not limited to this. For example,
The video described in the embodiment is not limited to a moving image, but includes a still image.

[0100]

As described above, according to the present invention,
Since the video with the embedded video feature is registered in the video database, and the video with the video feature embedded is searched and output from the video database, the video feature and the video are linked. Thus, it is possible to solve the problem of the prior art adopting a configuration of outputting in a communication data format.

In the present invention, since non-verbal search information of embedding video feature information is used,
Higher precision search can be expected compared to a search based on linguistic code information such as keywords.

In the present invention, when extracting features of a video, a combination of a plurality of algorithms is used to use a combination of remarkable characteristics of the video as search ID information. The processing load on the server side can be reduced as compared with a method of performing feature extraction such as performing semantic scene description in recognition.

In the present invention, a plurality of watermark embedding schemes having different algorithms are provided, and an algorithm suitable for the video feature quantity to be embedded or the video to be embedded is selected from among them, and the watermark information is used by using the selected algorithm. By embedding, the resistance can be increased.

In the present invention, since the database is stored in a form in which the image feature amount is directly embedded in the image, not only the database management is greatly simplified, but also the mismatch between the image and the search ID information during operation. Accompanying obstacles
Loss can be avoided.

[Brief description of the drawings]

FIG. 1 is an embodiment of the present invention.

FIG. 2 is an example of an embodiment of a database.

FIG. 3 is an explanatory diagram of a video scene.

FIG. 4 is an explanatory diagram of video features.

FIG. 5 is an explanatory diagram of a control signal issued by a database control unit.

FIG. 6 is an explanatory diagram of the present invention.

FIG. 7 is an explanatory diagram of a conventional technique.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Search system 2 Terminal device 10 HP part 11 Database 20 HP access part 21 Downloading correspondence part 100 GUI 101 Browsing video accumulating part 102 Receiving part 103 Video transmitting part 110 Video content accumulating part 111 Input interface part 112 Output interface part 113 Utility part

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) H04N 5/76 H04N 5/76 B 5/91 7/173 640A 7/173 640 5/91 N (72) Inventor Takao Nakamura 2-3-1 Otemachi, Chiyoda-ku, Tokyo Nippon Telegraph and Telephone Corporation (72) Inventor Noboru Sonehara 2-3-1 Otemachi, Chiyoda-ku, Tokyo Nippon Telegraph and Telephone Corporation ( 72) Inventor Hiroshi Sakai 1-1-12 Dogenzaka, Shibuya-ku, Tokyo NTT Electronics Corporation (72) Inventor Tomio Kishimoto 1-1-12 Dogenzaka, Shibuya-ku, Tokyo Inside NTT Electronics Corporation (72 Inventor Tsuneo Okubo 1-12-1 Dogenzaka, Shibuya-ku, Tokyo NTT Electronics Corporation F-term ) 5B075 ND06 ND12 NK06 NK21 NR02 PR06 QM08 5B082 EA07 5C052 AC08 DD04 5C053 FA07 HA29 JA21 5C064 BA07 BB10 BC18 BC20 BD02 BD08 BD09

Claims (8)

[Claims] A step of extracting a video feature from a video to be registered; a step of embedding the video feature in a video to be registered by using a digital watermark; Registering the video information in the video content storage unit while associating with the embedded video feature amount. 2. A first process of extracting a video feature from a video to be registered using a plurality of extraction algorithms, and a correspondence between the video feature and identification information of an extraction algorithm used for extracting the video feature. A second process of embedding in the video to be registered using a digital watermark while capturing the video with the video feature amount in the video content storage unit while associating the video with the embedded video feature amount. And a third step of registering the video information. 3. The video information registration method according to claim 2, wherein in the first step, when a video is composed of a plurality of frames, video feature amounts extracted from each frame are systematically arranged. A video information registration method characterized by processing to extract an object as a final video feature amount. 4. The video information registration method according to claim 2, wherein the management information of the video to be registered is embedded in the video to be registered using a digital watermark in the second step. A video information registration method characterized by processing. 5. The video information registration method according to claim 2, wherein in the second step, a plurality of digital watermark embedding algorithms are prepared, and an optimum one is selected from among them. A video information registration method characterized by selecting and using it to process the video feature amount so as to be embedded in the video to be registered. 6. A means for extracting a video feature quantity from a video to be registered by using a plurality of extraction algorithms, and associating the video feature quantity with identification information of an extraction algorithm used for extracting the video feature quantity. Means for embedding in a video to be registered using a digital watermark, and means for registering the video in which the video feature is embedded in the video content storage unit while associating the video with the embedded video feature. A video information registration device, characterized in that: 7. A step of inputting a browsing video with an embedded video feature using a digital watermark, a step of detecting a video feature embedded in the input browsing video, and a step of detecting the video feature. By accessing the video content storage unit that stores the embedded video while associating the video with the embedded video feature amount, the detected video is stored in the video content amount stored in the video content storage unit. A step of searching for a feature similar to the feature, and a step of reading and outputting, from the video content storage unit, a video in which the video feature associated with the searched video feature is embedded. Video information search method. 8. A means for inputting a browsing video with a video feature embedded using a digital watermark, a means for detecting a video feature embedded in the input browsing video, By accessing the video content storage unit that stores the embedded video while associating the video with the embedded video feature amount, the detected video is stored in the video content amount stored in the video content storage unit. Means for searching for a feature similar to the feature, and means for reading and outputting, from the video content storage unit, a video in which the video feature associated with the searched video feature is embedded. Video information search device.