Research On Motion Feature Extraction Method In Digital Video

It is generally thought of that visual content of video data composes of color, texture, shape, motion and so on. Among them, motion is peculiar content that it differs from image data. To a video clip, motion information reflects the changes which are the important information of the gradual progress in video .So if we want to describe the overall content of the video, the motion information is an essential respect, and it also is an essential part that carrying on motion analysis in video compression and content-based video retrieval.This paper discussed the feature extraction method of the movement object in video. It briefly introduced the related visual description content in MPEG - 7 standard first. On the basis of analyzing and comparing existing methods, it presents a motion feature extraction method which is based on movement object in video. The main thought of this method follows as below: it get the trajectory of the movement object in video by difference analysis firstly, and it considers the trajectory curvature as one of features of motion; then calculates the statistics feature of CCV of rectangle area containing movement object along the trajectory that it has formed, and this feature is considered as the second. At the same time, it proposes a video clip detection method which is based on the trajectory dots sequence order match. The characteristic of this method lies in simple computation and the great relativity between motion feature and the content of video, so it can satisfy the needs of content-based video query of different level users well.we have carried on the simulation experiment to the algorithms that are put forward to a segment of video, and has got the anticipated result.