Research & Realize Of CVR System

With the fast development of Internet and multimedia technology, there are more and more video resources which are widely used everywhere. Scholars and users have paid more and more attention to the research of the video information including much lively and abundant content. Content-based Video Analysis and Retrieval technology is one of the hot topics.The thesis focuses on Content-based Video Analysis and Retrieval technology. The main tasks and achievements are as follows:First, the RGB color feature, the HIS space feature and the HSV histogram feature are merged to describe the video frame's feature, which can be used to detect the shot cut more correctly.Second, the two-order difference of the adjacent frame is used to detect shot boundaries, which can avoid the influence of the camera motion in the shot.Third, an improved K-Means cluster algorithm is used to determine the key frames. The initial clustering centers are decided by the distribution of the video data which will make the result more stable and the selected key frames more representative.Fourth, the Euclidean distance is used to measure the similarity of the frames in my system. It can return the best matching videos for the input video.Fifth, a Video Analyze and Retrieval System is designed and implemented using Visual C++2008. The author has done many experiments using this system. The experiments show that the proposed method in this thesis is feasible and effective.Although the research of the thesis is junior and simple, the methods above have an important significance on the further research in this field. The system's performance can be improved by picking up more features and improving the algorithms.