KR200217878Y1 - Video editing, searching equipment - Google Patents

Abstract

The present invention facilitates the selection of algorithms for radical scene transformation and progressive scene transformation of various video sources, and maintains the original structure when creating a database for one video source. The present invention relates to a video editing and retrieval apparatus that can be managed in a database by efficiently managing them in a database.

The present invention for this purpose is to detect the transitions suitable for a variety of content from the AVI and MPEG files to convert the video stored in the video camera, video tape or direct video file to a video file through the capture device and to process the video file Algorithm integrator that integrates multiple algorithms for processing, setting and measuring unit for setting threshold and color histogram and camera information for each video, and detecting wrongly extracted scene frame key frame A scene transition detection, insertion, and deletion unit for deleting keyframes and inserting keyframes not obtained; a video indexing unit for inserting and indexing contents such as date, field, detail, and captions from extracted keyframes; Each video that is indexed is the original video without change A search browser that provides a browser that suits the user's taste by selecting a database from the logical database unit and logically organizing the structure between frames of each scene by logically databaseting only the information of the frames. It is characterized by the configuration.

Description

Video editing, searching equipment

The present invention relates to a video editing, retrieval apparatus.

Recently, due to the development of hardware and software and user's demand, multimedia technology has been actively improved. In particular, video is a very demanding database for storing, managing, and retrieving data due to its own huge data, and research on the construction of a content-based multimedia database is being actively conducted.

There are quite a few methods for algorithms for scene transitions. Histogram (HISTOGRAM) method using image (frame: minimum unit of video material) and image difference, method using edge, method using wavelet, method using pixel, etc. The method used is the color histogram. In the color model, the CMYK color model, the YIQ color model, and the HSI color model are used in the general R, G, and B color models.

In the method using a color histogram, there are many differences depending on each source video, it is very sensitive to the movement and noise of an object, and has a very difficult problem in applying a threshold. A database is essential for storing a lot of video data. However, in the related art, video data is separated into files and the linkage of each file is complicated.

In particular, the methods for segmenting a scene (sequence of consecutive tits of one main object) have a problem that it is difficult to extract a scene segment frame from various video files on a problem domain using only one algorithm. There is a situation.

The purpose of the present invention is to integrate various algorithms for efficiently constructing a video editing device and to provide instant and frequent threshold setting, easy insertion and deletion of scene frames, construction of a logical database, and retrieval of video according to user's taste. It is.

The present invention facilitates the selection of algorithms for the radical scene transformation and progressive scene transformation of various video sources to build a video editing device, and the original source is intact when creating a database for one video source. The technical challenge is to make only the logical structure maintained in a database so that it can be efficiently managed in terms of utilization and structure.

The present invention for achieving the above object is, first, select and use the most efficient algorithm by applying a number of presented algorithms. The threshold for algorithm selection can be manually modified and applied immediately. In addition, due to the limitation of the algorithm, the limitation of the existing algorithm is solved by using an additional insertion of a scene frame that is not extracted and a deletion function of an incorrectly extracted scene frame. Second, we extract the motion vector and affine parameters of the video to measure the dissolving and wipe scenes that are problematic in progressive scene segmentation. The camera analysis technique applied to segmentation is introduced. Third, a video database device was constructed by generating a logical pointer without modifying the video image of the original source, and presented a search device according to a user query. Fourthly, it provides a function to create and manage a user's specific file with a bundle of logical pointers to request scenes. It is characterized by.

The present invention effectively increases the efficiency of the scene segmentation and retrieval apparatus of the video through the convenience of the video editing apparatus and the response to the user's needs and tastes.

1 is an overall configuration diagram of a video editing and retrieval apparatus

2 is an integrated algorithm for a radical and gradual scene change method of video according to the present invention.

3 is a flow chart of insertion, deletion, threshold setting, and camera information measurement for video editing according to the present invention.

4 is a flowchart of a user's search and logical database construction according to the present invention.

<Description of the reference numerals for the main parts of the drawings>

10: Video input device 20: AVI and MPEG video

30: Algorithm integration unit for scene change

40: threshold setting, color histogram and camera measurement unit

50: scene change detection, insertion, deletion of extracted keyframes

60: video indexing unit

70: logical database portion

80: search browser 90: users

The present invention is designed to (1) integrate several algorithms suitable for various video moving images, (2) measure camera motion of motion vectors and affine coefficients, and (3) to detect falsely detected scene frames to overcome errors in extracted scene transform frames. To delete, insert scene frames not detected, (4) to build a logical database of the transition frame and annotation without changing the original video movie, (5) to provide a video search according to the user's taste effectively.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

1 is an overall configuration diagram of a video editing and retrieval apparatus according to the present invention, FIG. 2 is an integrated algorithm for detecting a frame of each video moving image and a progressive scene change and a radical scene change, and FIG. 3 is extracted. 4 is a flowchart for inserting, deleting a scene change frame, camera movement, affine parameters, and threshold correction. FIG. 4 is a flowchart for searching and building a logical database of a user according to the present invention.

1 is a view for explaining the overall operation of the video editing and retrieval device, an AVI and MPEG file that converts the video 10 stored from a video camera, video tape or direct video file to a video file through the capture device and processing 20, an algorithm integrator 30 for integrating various algorithms to detect scene transitions suitable for various contents from the stored video image file, and setting of threshold values and color histograms and camera information for each video. Scene to detect the setting and measurement unit 40 to measure and to detect a key frame (key frame (representative frame representing a tit)) which has been switched from the obtained information, to delete a key frame which is incorrectly extracted, and to insert a key frame not obtained. Transform detection, insertion and deletion unit 50, date (year, month, day, hour), field (de) from extracted key frames The video indexing unit 60 inserts and indexes contents such as e.g. shows, news, comedy, sports), details, and captions (subtitles), and the original video video of each indexed item is logically divided into only the information of frames. Searching to provide a browser suitable for the user's taste through region selection and favorites from the logical database unit 70 and the logical database unit 70 which logically organicizes the structure between frames of the scenes. It consists of a browser 80. In the figure, the user 90 may search for a video video by using the search browser 80.

FIG. 2 is an integrated algorithm for detecting key frames through scene transitions received from various (news, drama, show, sports, comedy, etc.) video files according to the _{present invention.} , Check the similarity of I ₂ (31). The algorithm used to use the color histogram calculation 32 for the radical scene change detection 35 uses an algorithm using a product of CHI-SQUARE and NTSC, a template matching algorithm, and the like. Cut in the drawing refers to a scene change point corresponding to a boundary between a scene and a scene as a cut.

HSI color model is used for gradual scene change detection 36 and various functions according to H, S, and I are used. The following equation is used to detect gradual scene effects of FADE and dissolve.

That is, the scene transition detection 33 of the fade means that the closer the proportional value of the contrast for the two frames is closer to 1 (-1), the more likely it is to fade in (out). Is the difference between The absolute value of the contrast is shown. The formula (34) of the dissolve is as follows.

As closer to 1, the probability of dissolve is higher.

Indicates the difference in hue, saturation and brightness.

3 is a diagram for obtaining a scene change detection 45 and an insertion 48, deletion 49, correction of a threshold 46, and information measurement 44 of a camera for the present invention. That is, the motion vector of the camera is extracted (42) from the image input signal 41, and the time motion indexes (Ft-1) of the X and Y axes of the extracted motion vectors, and the time motion indexes of the X and Y axes ( After extracting the feature (43) between the Ft), the scene change detection 45 is performed through the inference 44 considering the operation of the camera, and the manual editing of the threshold for the algorithm selection is performed immediately. 47). The equation for extracting the motion vector 42 of the camera is as follows.

In addition, the formula for calculating the Affine coefficient is as follows.

Where a ₁ is the magnification factor about the x-axis, a ₂ is the rotation factor about the x-axis, a ₃ is the movement factor about the x-axis, a ₄ is the rotation factor about the y-axis, and a ₅ is the y-axis The magnification factor for a, a _6, is the coefficient of motion for the y axis. After a scene change is detected 45, a frame that is recognized as an error frame through editor's interaction through an extracted key frame can be deleted 49 and a key frame could not be extracted due to an algorithmic problem. In this case, since the keyframe can be directly extracted through the insertion 48, the existing keyframe can not be found or the problem of the keyframe found by the error can be eliminated.

4 is a video indexing and retrieval apparatus. After completing the editing process of the transition from the video movie 71, each scene 72 has a database 73 of start and end frames 74 of the scene, which is logically structured without compromising the file of the video. In the conventional method, the video congestion and the database are complicated in the form of dividing one video into many files and serving each file to a user. In order to solve this problem, in the present invention, the information of each scene, which is one large video image file, is kept as it is. When (76) was present, the scenes for each query were created (77) in a new database. That is, after the user finds scenes that match the query 78 from various video files, they can be made into a new database and generated (77) into a video according to their taste. In this process, the database is constructed by logically connecting only the scenes of each video while maintaining the multiple video images.

As described above, the video editing and retrieval apparatus of the present invention will realize the automation of the broadcasting industry that targets images and videos, and furthermore, communication services such as video-on-demand, Internet TV, and Internet movies based on a multimedia database. Will stimulate research in the field.

Claims (1)

AVI and MPEG files 20 converting and storing video 10 stored from video cameras, video tapes, or direct video files into video files through capture devices, and scene changes suitable for various contents are detected from the stored video video files. Algorithm integration unit 30 for integrating and processing various algorithms, setting and measuring unit 40 for measuring threshold values and color histograms and camera information for each video, and scene-transformed keys from the obtained information. Scene transition detection, insertion, and deletion unit 50 for detecting frames, deleting wrongly extracted key frames, and inserting key frames not obtained, and contents of date, field, details, and captions from extracted key frames. The video indexing unit 60 inserts and indexes the indexed video, and the original video image of each of the indexed items is not changed. Logical database unit 70 which logically database only the information of frames to logically organize the structure between frames of scenes, and a browser suitable for a user's taste through selection and favorites from the logical database unit 70. A video editing and retrieval apparatus comprising a search browser (80) providing a.