Research On Video Retrieval Techniques Based On Motion Features

With the rapid development of computer, multimedia, and Internet techniques, Content-Based Video Retrieval (CBVR) becomes one of the most active research focuses of implementation of multimedia.In this paper, the background, the meanings of CBVR, the current states and active research areas are introduced, and the classical Content-Based Video Retrieval System (CBVRS) are presented, and then a brief review of the system frame and some important techniques in CBVRS are introduced.By using the technologies of image processing, video analysis, pattern recognition, computer vision and data mining, this dissertation studies some key problems in the field of video retrieval. The major contents include:The characteristics and the key techniques of the content-based video retrieval are analyzed systematically. The techniques and the functions of CBVR which are needed to be researched and consummated particularly were discussed fully.The motion features about video are studied deeply. The characteristics of the video motion features and the relevant techniques were analyzed in detail. Then, the functional modules and the retrieval flow based on the motion features of video were designed.An approach of video retrieval using video slice technique was proposed. This approach applies the video slice technique to compute and show the similarity between the shots. This technique was used in the field of video segmentation usually. The result of the experiment shows that this algorithm is effective.A video retrieval technique based on the trajectory of the local moving object was proposed. By combining the static features with the motion features of the local moving object, we proposed an algorithm about the local motion feature extraction by using the Camshift algorithm as the core. Then we design the integrate retrieval flow according to the characteristics of the CBVR and get the preferable result in the experiment.