Video Of Multiple Moving Targets, Space-time Integration Of The Segmentation Technique

Today the requirement on information is becoming the main motivation to promote information technology. Video information and its processing technology have made much progress. Because of the enormous data, it is quite difficult to be stored and lively transported. So the problem is hindering the application of digital video information. Hence it is urgent to study efficient representation of video data.In video sequences, we are interested in motion objects. So most of video segmentation refers to motion objects. This paper also mainly addresses the technology.This paper discusses a segmentation technology of multiple motion objects in video sequences based on spatio-temporal joint. We establish its system model, discuss the realization methods and point out some important technologies concerned. First, we analysis and create general theoretical model of video segmentation. Secondly, based on the model, an automatic detection algorithm of motion-changed regions is proposed from the point of fuzzy clustering. Motion-changed regions are detected, using a fuzzy clustering rule to distinguish between the relative noisy and motion-changed regions in time-differential image after global motion compensation. At last, we discuss the generation of video objects. Through the spatial difference technology, we get edge features in current frame. Combing the motion-changed regions and edge features, we generate video objects. Here, we discuss detection of closed support area of targets with orthogonal multi-stage techniques based on interior pixels. With the minimum-distance criteria,the generation of video objects technology is developed. PC experiment shows that the technologies are able to extract video objects swiftly and efficiently.