Analysis And Research On Content-based Video Retrieval System

With the development of multimedia and network technique, there are more and more multimedia information, so it becomes more and more difficult to search interested material. Hence, it comes forth a new research in the information technology, i.e. Content-Based Video Retrieval (CBVR).CBVR means video retrieval from enormous video database through matching of patterns derived from video analysis and feature extraction to represent its content. Based on MPEG-7, we have developed such a kind of model in prototype. The main work and contents in this thesis include:1) Based on analysis the characteristics of video, a general video data model, compatible with MPEG-7 standard, is established. In our model, videos are represented in a hierarchical structure with five layers from frame, shot, shot group, scene to story unit, and key frames, etc.2) Analysis the varies shot transformation and the corresponding shot detection method, then proposed one kind of auto-adapted sliding window with double threshold value shot examination algorithm ,which can simultaneously examine shear and gradual shot changes, and solved well the beginning localization question of gradual shot changes.3) In the key frame selection, has analyzed the traditional key frame selection method, and analyzed the camera movements to the frame's influence, proposed one kind of key frame retrieval based on the camera movement classification.4) Discussed the importance of group division in the video retrieval, and proposed examing group method based on different similar rank. On the basis of shot group, clustering the repeated scene and obtaining the video frequency abstract.5) Take the example video, analysis the video retrieval pattern based on the shot and scene, and introduction correlation feedback theory, in the user intervention to the retrieved video, and finally obtained user satisfaction effect.