The Research Of Content Based Video Retrieval Technique

With the development of information techniques and computer networks,and the extensive apply of multimedia, we can easily get a large number of video information. How to retrieve them quickly and accurately effectively is so important. Under the situation,the Content Based Video Retrieval(CBVR) technique has gradually become one of the most urgent tasks.Firstly, the foundation of video retrieval was researched in the paper. The base of color and texture feature was detailed researched in the paper, and introduction some other methods simply. Use the color and texture to retrieval the image.Secondly, the algorithms of shot boundary detection were researched in the paper. The shot boundary detection is the first step of the video retrieval. The paper summing up the classical algorithm, and did some research into shot segmentation based on mutual information emphasis. On the base of in-depth study mutual information, the paper proposed a gray histogram and LOG shot Gradual Transition algorithm against the interference of the movement and flash. Several types video were detection by the algorithm. We can compare the algorithm with others, and test the detection efficiency of the advanced algorithm in shot boundary detection.Lastly, the technology of key frames extraction was researched depth. The exist algorithm of this filed were briefly analyzed. To combine the intention of the shot boundary detection, and we have used the mutual information, the paper proposed a mutual information key frames extraction algorithm based on image block. The experiment results show that the key frames were extraction can reflect the content of the shot with the algorithm of the paper.