|
With the production and accumulation of the video program, the number of programs in a sharp increase. If people want to find their own video clips of interest in all the programs, it becomes more and more difficult. So there is an urgent need for a video search technology, from a large number of video programming resources for their own interest for some programs, example: a shot in the football match Lens, an actor works. This is a video retrieval. The first cited existing video retrieval methods, and analyzes their advantages and disadvantages. Video retrieval methods mainly text-based information retrieval method, frame-based video retrieval methods, and content-based retrieval method. Text-based retrieval techniques described in the text of the video clips, the video file name, for example, notes, title, and other information related video search. Perfect text search technology development, there are a large number of retrieval algorithms available, so the method is simple, small amount of calculation, the search faster. But because of the content of the text information is limited, can not every fragment of the detailed descriptions of the video, it is impossible detailed search to all the video clips to be examined, that is, search accuracy is low. A frame-based video retrieval, the reference frame is to be retrieved, and in each frame in each video comparison, can identify the requested video clip. However, since there are a large number of frames in the video, compare each frame is impossible, therefore appears a structured video analysis and retrieval, the video is divided into the video, scene, group, lens, five levels of the key frames, and sequentially from the top to low-level search, thus improving the search speed. But this is to use structural analysis techniques to video scene segmentation shot segmentation, compute-intensive than the complexity of the algorithm, the search speed is low. Content-based video retrieval, first, the video object segmentation, extraction and describes the features and contents of the video object stored in the object database. And then according to the characteristics of the object to provide the retrieval, in the object database retrieval. Its advantage is found full rate, the disadvantage is that the computational complexity. Of the above are several methods proposed subtitle-based video retrieval method. This method not only text retrieval method is simple and quick, but also has the precise characteristics of content retrieval method based lookup. The method first extracts the caption information in the video, and stored in the database. Just enter a caption information retrieval can retrieve the video appears in the subtitle.
|