Research On Key Frame Extraction In Content-Based Video Retrieval

With the rapid development of information technology and internet technology, multimedia information data especial video data has become increasingly rich and become an important method to get information and exchange. However, because data volume is huge and video data is unstructured, and the video content is rich, it is difficult for video analyzing and video processing. How to use and manage multimedia data effectively is becoming the issue people increasingly concerned about.Traditional video retrieval based on text annotation need people annotate video manually. More and more video information are available, the manual annotation is time-consuming and inconvenient. In addition, video content is rich, manual annotation can’t describe the content comprehensively and accurately. Furthermore, manual annotation is subjective. All these make the influence on video retrieval directly. Content based video retrieval technology has gradually become the hot issue of current research at home and abroad.Content based video retrieval extracts the low level features(color, texture, shape) and high semantic feature and then establishes the video index. Provide users with different query methods and analyze video submitted by users automatically. Then similarity comparison is done between video submitted by users and video in data base. At last, results are presented according to the similarity. Content based video retrieval includes many key technologies such as shot detection, key frame extraction, feature extraction, similarity comparison and query methods. Key frame extraction technology is an important step in video retrieval. This paper mainly aims at the study and realization of key frame extraction algorithm.Firstly, this paper introduces the research background and significance of video retrieval technology and the overseas and domestic research status.Next, by reading a lot of literature, content based video retrieval technology is described in detail especial key frame extraction.Then, several effective key frame extraction methods are proposed:key frame extraction based on frame blocks differential accumulation; key frame extraction based on video spatio-temporal slice. And a simple key frame extraction system is designed by VC++and OpenCV.At last, this paper summarizes the development direction of video retrieval and key frame extraction.