JP2009171018A - Summary video image creation apparatus, summary video image creation method, summary video image creation program, picture analyzer, picture analysis method, and picture analysis program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To make a summary without missing a scene important in a program in continuous video image data of a drama, or the like. <P>SOLUTION: In a summary video image creation apparatus, an outline video image extraction section extracts "the last outline video image" from a future video image after a video image becoming an object for making a summary. Consequently, a summary of the entire story including the highlights of each story, which cannot be attained in conventional summary made only from the next trailer, can be made. When "the last outline video image" is extracted, an outline video image discrimination section can extract the n-th story as well as only the scenes which actually feature the n-th story while removing flashbacks and scenes common to respective stories with reference to previous stories. With this feature, detection accuracy of "the last outline video image" is enhanced and an exact summary can be made. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a summary creation device, a summary creation method, a summary creation program, and a video analysis device that analyzes video content, a video analysis method, and a video analysis program.

In recent years, with the start of digital broadcasting and the widespread use of recording devices such as video equipment that records and plays back video data, summaries of each story are automatically created and viewed in programs composed of multiple episodes such as TV dramas. Thus, a method and an apparatus for grasping the contents of a rough video have been proposed. For example, as a conventional summary video creation technique, there is a technique that uses that an important scene in the nth broadcast of a continuous drama video is used as a next notice in the n-1th broadcast (for example, , See Patent Document 1). That is, the (n-1) -th next notice is extracted, and the n-th program summary is automatically created based on the information. This makes it possible to create a summary.
JP 2003-163881 A

However, the technique disclosed in Patent Document 1 has a problem that the summary is often incomplete because the summary of the program is created only from the previous previous notice. Usually, the next notice is an introductory part for attracting viewers and does not include the final part of each story and the conclusion part. Therefore, the summary created by the technique disclosed in Patent Document 1 often includes only the first half of the program and does not include the core part of the program.
The present invention has been made in view of the above points, and an object of the present invention is to create a summary without missing an important scene as a program of each story in continuous video data such as a drama.

  The present invention has been made in order to solve the above-mentioned problems. In a program composed of a plurality of episodes such as dramas, the previous rough stripes and climax are often inserted before the main story of n + 1 episodes. This property is used. According to one aspect of the present invention, in a summary video creation device that generates a summary of video data, the video data of the (n + 1) th episode is compared with the video data of the (n + 1) th episode, and the video data of the (n + 1) th episode is compared with the video data of the nth episode. A rough muscle image extracting unit for extracting “previous coarse muscle video” as data, and a summary video creating unit for generating summary data of the nth episode using the extracted “previous rough muscle video”. This is a featured summary video creation device.

  According to the present invention, in the summary video generation device, the “previous coarse muscle video” is acquired from the future video rather than the video for which the summary is to be created by the coarse muscle video extraction unit. As a result, it is possible to create an overall summary including the climax of each story, which was lacking when creating a summary only from the previous next notice.

  Further, according to one aspect of the present invention, in the summary video creation device, the extracted “previous coarse video” is compared with video data of each story before the nth episode, and the “ A rough muscle video discriminating unit for discriminating only “previous coarse muscle video” unique to the n-th episode from the “previous coarse muscle video”, and the summary video creation unit includes the extracted “previous coarse muscle video” Only the “previous coarse video” unique to the nth episode is used.

  According to the present invention, when extracting the “previous coarse video”, the coarse video discriminating unit refers not only to the nth episode but also each episode before it, and is used as a recollection scene or a guide before the program starts. It is possible to remove only scenes that are common to each story, such as catch, and to extract only the scenes that are actually the features of the nth episode. This feature improves the accuracy of detection of the “previous coarse image” and enables accurate summarization.

  According to another aspect of the present invention, in the summary video creation device, the summary video creation unit generates summary data of the nth episode using the preview video extracted from the video data of the n-1th episode. It is characterized by that.

  According to the present invention, in the summary video generation device, the summary video creation unit uses the “preliminary video” acquired from the past video in addition to the “previous coarse video”. As a result, in addition to the climax of each story, it is possible to create a more accurate summary including an introductory part for attracting viewers.

  Further, according to one aspect of the present invention, in the above summary video creation device, the coarse line video extraction unit determines the length of the “previous coarse line video” extracted continuously, and is not unique to the nth episode. It is characterized by removing.

  According to the present invention, images that are not unique to the nth episode, such as (1) the theme song and (2) a momentary black screen at the time of a scene change, are removed, and only the video unique to the nth episode is accurately extracted. Can do.

  Further, according to one aspect of the present invention, in the above summary video creation device, the coarse video extraction unit extracts a CM at the beginning of the (n + 1) -th episode, and the “previous coarse video” from immediately after the extracted CM. The extraction process is started.

  According to this invention, since the “previous coarse image” is arranged at the beginning of the program, the extraction process of the “previous coarse image” is started immediately after the CM using the feature that the CM is immediately before the program. Therefore, the “previous coarse image” can be found in a short time.

  In addition, according to one aspect of the present invention, in a video analysis apparatus that analyzes video content, the video data of the (n + 1) th episode is compared with the video data of the (n + 1) th episode, and the video data of the (n + 1) th episode is compared with the video data of the nth episode. The “previous coarse muscle video” is extracted, and the extracted “previous coarse muscle video” is compared with the video data of each story before the nth episode. An image analysis apparatus comprising: a coarse muscle image discriminating unit that discriminates only "previous coarse muscle image" unique to the nth episode from "previous coarse muscle image".

  According to another aspect of the present invention, in a summary video creation method for generating a summary of video data, the video data of the (n + 1) th episode is compared with the video data of the (n + 1) th episode, and the video data of the (n + 1) th episode is compared with the nth episode. A rough muscle video extracting step for extracting “previous coarse muscle video”, which is video data of the above, and a summary video creating step for generating summary data of the nth episode using the extracted “previous rough muscle video” This is a method for creating a summary video.

  Further, according to an aspect of the present invention, in the above summary video creation method, the extracted “previous coarse muscle video” is compared with video data of each story before the nth episode, and the “ A rough muscle image discriminating step for discriminating only the “previous coarse muscle image” unique to the n-th episode from the “previous coarse muscle image” is included. Only the “previous coarse image” unique to the nth episode is used.

  According to another aspect of the present invention, in the summary video creation method, the summary video creation step generates summary data of the nth episode using the preview video extracted from the video data of the n-1th episode. It is characterized by that.

 Further, according to one aspect of the present invention, the length of the “previous coarse muscle video” continuously extracted in the coarse video extraction step of the summary video creation method is determined, and the video that is not unique to the nth episode is determined. It is characterized by removing.

  Also, according to one aspect of the present invention, in the coarse video extraction step of the summary video creation method described above, a step of extracting a CM at the beginning of the (n + 1) -th episode, and extraction of “previous coarse video” immediately after the extracted CM And a step of starting processing.

  According to another aspect of the present invention, in the video analysis method for analyzing video content, the video data of the (n + 1) th episode is compared with the video data of the (n + 1) th episode, and the video data of the (n + 1) th episode is compared with the video data of the nth episode. The “previous coarse muscle video” is extracted, and the extracted “previous coarse muscle video” is compared with video data of each story before the nth episode. And a rough muscle image discriminating step for discriminating only the “previous coarse muscle image” unique to the nth episode from the “previous coarse muscle image”.

  One embodiment of the present invention is a program for generating a summary of video data, comparing video data of the (n + 1) th episode with video data of the (n + 1) th episode, and from the video data of the (n + 1) th episode to the nth video data. A rough muscle video extraction step of extracting “previous coarse muscle video” which is video data of the story, and a summary video creation step of generating summary data of the nth episode using the extracted “previous rough muscle video”. A program for causing a computer to execute.

  Further, according to one aspect of the present invention, in the above program, the extracted “previous coarse image” is compared with video data of each episode before the nth episode, and the “previous coarse line” is compared. A rough muscle video discriminating step for discriminating only the “previous coarse muscle video” unique to the n-th episode from the “video”, and the summary video creating step includes the nth of the extracted “previous coarse muscle video”. Only the “previous coarse muscle video” unique to the story is used.

  Further, according to one aspect of the present invention, in the above program, the summary video creation step generates summary data of the nth episode using a preview video extracted from the video data of the n-1th episode. And

 Further, according to one aspect of the present invention, in the coarse line image extraction step of the program described above, the length of the “previous coarse line image” continuously extracted is determined, and an image that is not unique to the nth episode is removed. It is characterized by that.

  Further, according to one aspect of the present invention, in the coarse-muscle image extraction step of the above program, a step of extracting a CM at the beginning of the (n + 1) -th episode and a process of extracting “previous coarse-video image” immediately after the extracted CM And a step of performing.

  One embodiment of the present invention is a computer program for analyzing video content, comparing video data of the (n + 1) th episode with video data of the (n + 1) th episode, and from the video data of the (n + 1) th episode to the nth episode. The rough muscle image extraction step for extracting the “previous coarse muscle video” which is the video data of the video, and the extracted “previous coarse muscle video” are compared with the video data of each story before the nth episode. , A program for causing a computer to execute a coarse image determining step for determining only the “previous coarse image” unique to the nth episode from the “previous coarse image”.

According to the present invention, in the summary video generation device, the “previous coarse muscle video” is acquired from the future video than the video for which the summary is to be created by the coarse video extraction unit. As a result, it is possible to create an overall summary including the climax of each story, which was lacking when creating a summary only from the previous next notice.
In addition, according to the present invention, when extracting the “previous coarse image”, the coarse image discriminating unit refers not only to the nth episode but also the previous episodes, and is used for the recollection scene and each episode. The scene is removed, and only the scene that is actually the feature of the nth episode can be extracted. This feature improves the accuracy of detection of the “previous coarse image” and enables accurate summarization.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a block diagram showing a configuration of a summary creation device according to an embodiment of the present invention.
The summary creation device 1 includes a video reception unit 101, a video storage unit 102, a CM detection unit 103, a coarse muscle video extraction unit 104, a candidate scene storage unit 105, a coarse video discrimination unit 106, and a summary video creation unit 107. And a display unit 108 that displays the video to the viewer. The summary generation device 1 acquires the “previous coarse line video” from the video data of the (n + 1) -th episode in response to an explicit input from the user or a preset process start timing of the program summary. A processing function for creating a summary video of the nth episode based on the “previous coarse video” is provided. The “previous coarse image” is a scene in which the previous scene and the outline so far are reviewed at the beginning of a drama or the like.

The video receiving unit 101 receives video data of a program through digital TV broadcasting, IP (Internet Protocol) communication, or the like. Note that video data is composed of a plurality of frames. The video storage unit 102 stores video data of programs received by the video receiving unit 101 in accordance with reservation information and user preferences. In addition, it is assumed that the number of episodes corresponding to the stored program is obtained based on EPG (Electronic Program Guide) information at the time of program recording reservation.
The CM detection unit 103 detects CM (commercial) data from the video data. As a method for detecting CM data, for example, a method as disclosed in JP-A-8-339379 is used.

The coarse muscle video extraction unit 104 acquires video data of the (n + 1) th episode from the video storage unit 102, and extracts “previous coarse muscle video” from the acquired video data of the (n + 1) th episode.
The candidate scene storage unit 105 stores the “previous coarse image” extracted by the coarse line image extraction unit 104 as a “previous coarse image” candidate. Candidate scene storage unit 105 may store it in its own real memory or store video in video storage unit 102.

The coarse muscle video discriminating unit 106 acquires the “previous coarse muscle video” candidate extracted from the video data of the (n + 1) th episode from the candidate scene storage unit 105, and the acquired “previous coarse muscle video” candidate and the previous episode of the nth episode. By comparing with the video of each story, the “previous coarse muscle video” appropriate as the coarse muscle of the nth episode is extracted.
The summary video creation unit 107 creates a summary video (summary data) of the nth episode based on the “previous coarse muscle video” extracted by the rough muscle video discrimination unit 106.

FIG. 2 is a flowchart illustrating an example of a processing flow in the coarse line video extraction unit 104. The “previous coarse video” is arranged at the beginning of the program due to its nature. Moreover, the following cases can be considered as a general program structure.
(A) CM → “Previous coarse muscle image” → Theme song → CM → Main part (b) CM → Theme song → CM → “Previous coarse line image” → Main part (c) CM → “Previous coarse line image” → Main part ( d) CM-> theme song->CM-> main story (program without "previous coarse video")
(E) CM → main part (program without “previous coarse video”)
In the cases (a) to (c) where the “previous coarse image” is entered, the “previous coarse image” is always included after the CM. Therefore, in the present invention, the CM at the beginning of the recorded video data is detected, and “previous coarse video” is extracted from the data immediately after that.

  Here, the case where the “previous coarse image” of the (n + 1) th episode is extracted will be described as an example. The coarse line video extraction unit 104 acquires the video data of the (n + 1) th episode from the video storage unit 102 in advance, and extracts the first frame of the acquired video data. The number of frames of the video extracted here is determined according to the length of the entire video (for example, the number of frames corresponding to the first 10 minutes for a 60-minute drama). Next, the coarse line image extraction unit 104 acquires the start point and end point of the CM data in the extracted first frame from the CM detection unit 103.

The coarse line image extraction unit 104 starts detecting a matching frame from the end point of the CM data (step S1). Here, the matching frame is a frame that exists in the nth story by comparing the extracted first frame with each frame in the nth story. At this time, the coarse line image extraction unit 104 ends the detection when no matching frame is detected for a certain period of time (a predetermined threshold t ₁ second) from the start of detection (step S2). This is because it is determined that this is not the “previous coarse line image”. Further, when the coarse line image extraction unit 104 reaches the start point of the CM data detected by the CM detection unit 103 during the detection (step S3), the detection ends. In this case, it is determined that the “previous coarse image” has ended.

  The coarse line image extraction unit 104 determines whether or not the matching frames detected by the above-described steps are continuous (step S4). If not continuous (step S4: No), the coarse line image extraction unit 104 moves to the next frame and performs the process of step S2.

When there is a continuous frame (step S4: Yes), the coarse line image extraction unit 104 determines whether or not the frames continuously match for a predetermined threshold t ₂ seconds or more (step S5). Frames that do not match continuously for t ₂ seconds or more are not adopted as the “previous coarse line image” in order to remove the black image at the time of scene change or video connection. In that case (step S5: No), the coarse line image extraction unit 104 moves to the next frame and performs the process of step S3. Here, t ₂ is a very short threshold, and is assumed to be about several hundred milliseconds.
In addition, the coarse line image extraction unit 104 determines whether or not the matched frames continue for a threshold value t ₃ seconds or more (step S6). The “previous coarse image” is a combination of the previous scenes, and each individual scene is very short. For this reason, if a video having a length of t ₃ seconds or more matches the previous video, it is assumed that it is a common part of each story such as a theme song, and is not adopted as the “previous coarse video”. In this case, the coarse line image extraction unit 104 moves to the next frame and performs the process of step S3. Here, the threshold t ₃ is assumed to be about several tens of seconds, and the relationship of t ₂ <t ₃ is established.

When the continuous frame is t ₂ seconds or more and less than t ₃ seconds, the video data composed of the continuous frames is adopted as the “previous coarse line video”. The adopted video data is stored in the candidate scene storage unit 105 as “previous coarse video” candidates (step S7).

The coarse line image extraction unit 104 repeats such a procedure and stores the “previous coarse line image” candidate. This process, in step S2, if a frame matching t least _one second was not detected (step S2: Yes) to exit. In this way, the coarse line video extraction unit 104 extracts “previous coarse line video” candidates. The number of “previous rough-line video” candidates to be stored is arbitrary.

FIG. 3 is a conceptual diagram illustrating an example of processing in the coarse line image determination unit 106.
In the prior art described in Japanese Patent Laid-Open No. 2003-163881, a summary is created using the next notice. In the next notice, it can be expected that all the images used in the next notice broadcast in the (n-1) -th episode are images of the n-th episode. On the other hand, in the “previous coarse image” used in the present invention, the image used in the “previous coarse image” of the (n + 1) -th episode is not necessarily the image of the n-th episode, but the image of the previous episode is also included. May be included. Also, even if the video appears in the nth episode, if it is a reminiscence scene of the previous scene, the scene does not represent the feature of the nth episode and is suitable for use in the summary video. Not. Therefore, in the present invention, after extracting the “previous coarse muscle video” candidate by the coarse muscle video extracting unit 104, the coarse muscle video discriminating unit 106 compares the video with each story before the nth episode. The scenes included in the “previous coarse image” are extracted. Here, the “previous coarse image” candidate is composed of a plurality of scenes, and each scene is composed of a plurality of frames.

  FIG. 3 shows an example in which the summary of the nth episode is generated from the “previous coarse image” of the (n + 1) th episode. It is assumed that scenes (1), (3), and (4) are extracted from the image of the (n + 1) -th episode as “previous coarse-line video” candidates. However, here, (1) is the scene of the n-2 episode, and the scene (1) 'in the nth episode is the recollection scene of the scene (1). In this case, the scene (1) 'is extracted as the rough streak of the nth episode by the rough streak video extraction unit 104, but actually the scene (1) is not a scene representing the feature of the nth episode. Accordingly, the coarse line video discriminating unit 106 compares each scene of the “previous coarse line video” candidate scene with the video of each story before the nth episode.

The coarse line video discriminating unit 106 acquires a “previous coarse line video” candidate from the candidate scene storage unit 105. The coarse line video discriminating unit 106 compares each acquired “previous coarse line video” candidate with past program data (each episode before the n-th episode). Here, the comparison method detects a matching frame for each scene. When the continuous matching frames are t ₂ seconds or more and less than t ₃ seconds, the coarse-line image discriminating unit 106 classifies the scene into “scenes appearing in the n-th story and previous stories”. The coarse line image discriminating unit 106 classifies the “previous coarse line image” candidates into “scenes appearing only in the nth episode” and “scenes appearing in the nth episode and the previous episodes” (in the example of FIG. 3) and (4) are “scenes that appear only in the nth episode”, and (1) are “scenes that appear in the nth episode and previous stories”). The rough muscle video discriminating unit 106 outputs “scene that appears only in the nth episode” to the summary video creating unit 107 as “previous coarse muscle video”.

The summary video creation unit 107 creates a summary video based on the “previous coarse muscle video” selected by the rough muscle video discrimination unit 106.
The length of the summary video to be created is determined by a user input or a fixed value of the system. The summary video creation unit 107 divides this length by the number of scenes extracted by the coarse line video discrimination unit 106 to determine how long one scene should be.

  The summary video creation unit 107 acquires a scene from the start point (or before and after) of each scene point extracted as the “previous coarse video” of the n + 1th episode of the main story of the nth episode (Step S1). S31). Specifically, a scene change is detected by extracting a video motion vector and a silent section, and the scene from the starting point to the detected scene change point is acquired as one scene. The summary video creation unit 107 compares the length of the detected scene with the determined length of one scene and, if the length is insufficient, repeatedly obtains the scene until the length is reached.

Next, the summary video creation unit 107 connects the acquired scenes to obtain a summary video (step S32). At this time, the summary video creation unit 107 adjusts by a fade process or the like when the length of the joined video and the length of the summary video determined in advance do not match.
In this way, the summary video creation unit 107 creates a summary video.

As described above, according to the present embodiment, in the summary video generation device 1, the coarse line video extraction unit 104 acquires “previous coarse line video” from a future video than the video for which the summary is to be created. As a result, it is possible to create an overall summary including the climax of each story, which was lacking when creating a summary only from the previous next notice.
In addition, when extracting the “previous coarse video”, the coarse video discriminating unit 108 refers to not only the nth episode but also the previous episodes, and removes the recollection scenes and the scenes common to the respective episodes. It is possible to extract only the scene that is the feature of the nth episode. This feature improves the detection accuracy of the “previous coarse line video” and enables accurate summarization.

Next, a summary video creation device 2 according to a second embodiment of the present invention will be described.
FIG. 4 is a block diagram showing a configuration of the summary creation device 2 according to the second embodiment. In the first embodiment, the summary video creation unit 107 creates a summary of the program using only the “previous coarse line video”, but in this embodiment, in addition to the “previous coarse line video”, the “preliminary video” To create a summary of the program. Therefore, the summary creation device 2 of the present embodiment further includes a preview video extraction unit 109. The reserved video extraction unit 109 extracts the “notice video” of the n−1th episode. Here, as a method for extracting the “preliminary video”, the conventional technique described in Japanese Patent Laid-Open No. 2003-163881 is used. Since other configurations are the same as those of the first embodiment, description thereof is omitted.

  FIG. 5 is a conceptual diagram illustrating an example of processing of the summary video creation unit 107 in the second embodiment. In this figure, when the summary video of the nth episode is created, the “preliminary video” of the (n−1) th episode is used in addition to the “previous coarse video” of the (n + 1) th episode. Both the scene extracted in the “notice video” (scenes A, B, and C in the example in the figure) and the scene extracted as “previous coarse video” (scenes (3) and (4) in the example in the figure) On the other hand, by performing the processing described in steps S31 and S32, a summary based on the entire nth episode is realized.

  As described above, according to the present embodiment, the summary video generation device 107 uses the “preliminary video” acquired from the past video in addition to the “previous coarse video” by the summary video creation unit 107. As a result, in addition to the climax of each story, it is possible to create a more accurate summary including an introductory part for attracting viewers.

  Further, a summary video is created by recording a program for realizing the function of the processing unit in FIG. 5 on a computer-readable recording medium, causing the computer system to read and execute the program recorded on the recording medium. Processing may be performed. Here, the “computer system” may include an OS and hardware such as peripheral devices.

Further, the “computer system” includes a homepage providing environment (or display environment) if a WWW system is used.
The “computer-readable recording medium” means a flexible disk, a magneto-optical disk, a ROM, a writable nonvolatile memory such as a flash memory, a portable medium such as a CD-ROM, a hard disk built in a computer system, etc. This is a storage device.

Further, the “computer-readable recording medium” means a volatile memory (for example, DRAM (Dynamic Random Access Memory)), etc., which hold programs for a certain period of time.
The program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line.
The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, and what is called a difference file (difference program) may be sufficient.

As described above, the embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to the above, and various design changes and the like can be made without departing from the scope of the present invention. It is possible to
For example, the processing performed by the coarse image extracting unit 104 and the coarse image determining unit 105 may be provided in a video analysis device or program that analyzes video content.

It is a block diagram which shows the structure of the summary production apparatus by one Embodiment of this invention. It is a flowchart showing an example of the flow of the process in the rough-line image | video extraction part by this embodiment. It is a conceptual diagram showing an example of the process in the coarse image | video discrimination | determination part by this embodiment. It is a block diagram which shows the structure of the summary production apparatus in 2nd embodiment. It is a conceptual diagram showing an example of the process of the summary image | video preparation part in 2nd embodiment.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Summary video production apparatus 2 ... Summary video production apparatus 101 ... Video reception part 102 ... Video storage part 103 ... CM detection part 104 ... Coarse muscle video extraction part 105 ... Candidate scene preservation | save part 106 ... Coarse muscle video discrimination | determination part 107 ... Summary video production Part 108 ... display part 109 ... notice video extraction part

Claims (18)

In a summary video creation device that generates a summary of video data,
A coarse-muscle video extraction unit that compares video data of the n + 1-th episode with video data of the n-th episode and extracts “previous coarse-line video” that is video data of the n-th episode from the video data of the n + 1-th episode;
Using the extracted “previous coarse muscle video”, a summary video creation unit that generates summary data of the nth episode;
A summary video creating apparatus comprising: The extracted “previous coarse muscle video” is compared with video data of each episode before the nth episode, and the “previous coarse muscle video” unique to the nth episode is selected from the “previous coarse muscle video”. It has a coarse line image discriminating unit that discriminates only `` video '',
The summary video creation unit uses only the “previous coarse muscle video” unique to the nth episode among the extracted “previous coarse muscle video”,
The summary video creation apparatus according to claim 1.   3. The summary video creation device according to claim 1, wherein the summary video creation unit generates summary data of the nth episode using a preview video extracted from the video data of the n−1th episode. .   The summary video according to claim 1, wherein the coarse image extraction unit determines the length of “previous coarse image” extracted continuously and removes an image that is not unique to the nth episode. Creation device.   2. The summary according to claim 1, wherein the rough muscle image extracting unit extracts a CM at the beginning of the (n + 1) -th episode and starts a process of extracting a “previous coarse image” immediately after the extracted CM. Video creation device. In a video analysis device that analyzes video content,
A coarse-muscle video extraction unit that compares video data of the n + 1-th episode with video data of the n-th episode and extracts “previous coarse-muscle video” that is the video data of the n-th episode from the video data of the n + 1-th episode;
The extracted “previous coarse muscle video” is compared with video data of each episode before the nth episode, and the “previous coarse muscle video” unique to the nth episode is selected from the “previous coarse muscle video”. A coarse line image discriminating unit for discriminating only “video”,
A video analysis apparatus comprising: In a summary video creation method for generating a summary of video data,
A coarse muscle video extraction step of comparing the video data of the n + 1th episode with the video data of the nth episode and extracting the “previous coarse muscle video” that is the video data of the nth episode from the video data of the n + 1th episode;
A summary video creation step of generating summary data of the nth episode using the extracted “previous coarse muscle video”. The extracted “previous coarse muscle video” is compared with video data of each episode before the nth episode, and the “previous coarse muscle video” unique to the nth episode is selected from the “previous coarse muscle video”. A coarse line image discrimination step for discriminating only "video",
The summary video creation step uses only the “previous coarse image” unique to the nth episode among the extracted “previous coarse image”.
The summary video creation method according to claim 7, wherein:   The summary video creation method according to claim 7 or 8, wherein the summary video creation step generates summary data of the nth episode using a preview video extracted from the video data of the (n-1) th episode. . In the coarse line image extraction step, the length of the “previous coarse line image” continuously extracted is determined, and an image that is not unique to the nth episode is removed.
The summary video creation method according to claim 7, wherein: In the rough muscle video extraction step of extracting “previous coarse video” which is video data of the nth episode from video data of the n + 1th episode,
Extracting the CM at the beginning of episode n + 1;
A step of starting extraction processing of “previous coarse image” immediately after the extracted CM;
The summary video creation method according to claim 7, further comprising: In a video analysis method for analyzing video content,
A coarse muscle video extraction step of comparing the video data of the n + 1th episode with the video data of the nth episode and extracting the “previous coarse muscle video” that is the video data of the nth episode from the video data of the n + 1th episode;
The extracted “previous coarse muscle video” is compared with video data of each episode before the nth episode, and the “previous coarse muscle video” unique to the nth episode is selected from the “previous coarse muscle video”. Coarse line image discrimination step for discriminating only "video",
A video analysis method characterized by comprising: A program for generating a summary of video data,
A coarse muscle video extraction step of comparing the video data of the n + 1th episode with the video data of the nth episode and extracting the “previous coarse muscle video” that is the video data of the nth episode from the video data of the n + 1th episode;
A summary video creation step of generating summary data of the nth episode using the extracted “previous coarse muscle video”;
A program that causes a computer to execute. The extracted “previous coarse muscle video” is compared with video data of each episode before the nth episode, and the “previous coarse muscle video” unique to the nth episode is selected from the “previous coarse muscle video”. A coarse line image discrimination step for discriminating only "video",
The summary video creation step uses only the “previous coarse image” unique to the nth episode among the extracted “previous coarse image”.
The program according to claim 13.   15. The program according to claim 13, wherein the summary video creation step generates summary data of the nth episode using a preview video extracted from the video data of the n-1th episode. In the coarse line image extraction step, the length of the “previous coarse line image” continuously extracted is determined, and an image that is not unique to the nth episode is removed.
The program according to claim 13. In the rough muscle video extraction step of extracting “previous coarse video” which is video data of the nth episode from video data of the n + 1th episode,
Extracting the CM at the beginning of episode n + 1;
A step of starting extraction processing of “previous coarse image” immediately after the extracted CM;
The program according to claim 13, comprising: A program for analyzing video content,
A coarse muscle video extraction step of comparing the video data of the n + 1th episode with the video data of the nth episode and extracting the “previous coarse muscle video” that is the video data of the nth episode from the video data of the n + 1th episode;
The extracted “previous coarse muscle video” is compared with video data of each episode before the nth episode, and the “previous coarse muscle video” unique to the nth episode is selected from the “previous coarse muscle video”. Coarse line image discrimination step for discriminating only "video",
A program that causes a computer to execute.

Images