Feature-based Video Retrieval In Compressed Domain

ABSTRACT:With the rapid development of multimedia computing and network transmission technology continues to improve, People can access to the rapid growth of multimedia data.Video is the most complex media form in Multimedia data, with its diverse forms, rich semantic content, and convenient recording.It has been widely applied and developed. How to analysis and retrieval video information quickly and effectively has become an urgent need to address an important issue. This paper proposes a method that combines the texture feature and the motion feature. This paper studies and achieves the methods of extracting the texture feature and motion feature from the compressed video and retrieving the video.This paper proposes a content-based video retrieval solution, it extracts the texture feature and motion feature directly and studies these features, introducts fuzzy decision theory to realize feature fusion and video retrieval.According to this algorithm, an application system of content-based video retrieval in compressed domain is designed and realized. The main contents of the paper are as follows:In this paper, firstly the background and meaning of this subject are given, and this paper introduces the development history and current situation of video retrieval at home and abroad. On this basis, the basic theory of video compression and MPEG-2 and MPEG-7 standards are introduced. In addition, content-based video retrieval status at home and abroad is also introducted.Secondly, the motheds of content-based video retrieval in compressed domain is studied carefully.The motheds of extracting texture feature and motion feature are analyzed. The paper proposeds feature fusion using fuzzy mathematics and the use of cosine angle made to measure the similarity between two video. On this basis, this paper proposeds a content-based video retrieval algorithm.Then, the programming implementation which aimed to the above algorithm is given, and a content-based video retrieval in compressed domain system is completed, and experimental results show that the system's output is in line with the expected results.Finally, a summary of the work of the full text and my own understanding of the development of video retrieval is given, what's more, it pointing out the struggling direction for future work.