Efficient Content-based Video Retrieval Technology

Recently, with the rapid development of the computer, network and communication technologies, the amount and types of information are booming over the Internet day and day. How to get the required information from this huge multimedia library is becoming a very important subject. This thesis is proposed according to this requirement and put emphasis on video. The researching content is about video structure analysis and the description principle of features.Firstly, introductions to the content of video structure analysis and the algorisms about it. According to the importance of shot segmentation in content-based retrieval or CBR, we have progressed the automatic threshold detection method. Experiments show that this progressed algorism can provide satisfactory shot segmentation results in common video data. Secondly, how to apply classic key-frame extraction methods to compact video have been proposed in this paper.Thirdly, MPEG-7 descriptors and description schemes are detailed given. Then, feature definition, representation and retrieval forms in MPEG-7 as well as its application in video retrieval are discussed.At present, Most of research in content-based video retrieval is in the theory part and there is no successful system to all kinds of video. My paper mainly focused on the theoretical research and a universal video retrieval system is devised.Finally, the application prospect and future trend of CBR are approved in the summarization of the whole thesis.