Research Of Human Motion Tracking Technology Based On Video

Video based human motion tracking technology is a hot issue in the field of computer vision. The main target is detecting, tracking human body, getting motion data, edit and use of the human data have been obtained, finally to allow the machine to understand and analyze the behavior of the human. In video surveillance, sports analysis, assistant clinical diagnosis, human animation, games, virtual reality and augmented reality, human-computer interaction and other fields, it has broad application prospects. Based on researching on various kinds of human motion tracking method, the works are introduced as two aspects as follows.1. Proposed an improved particle filter algorithm for human motion tracking.For the issues of accuracy and efficiency existing on the video based human motion tracking system, we presented an improved particle filer algorithm, because of difficulties in tracking on a single feature, in order to improve the tracking accuracy, we consider a variety of feature such as silhouette, edge, gradient etc to the filter; for the efficiency, we adopt the method of likelihood model based on observation and sampling in the gradient direction to optimize.2. Design and implementation of a prototype human motion tracking system based on improved particle filter algorithm.On the basis of the above improved particle algorithm, we realized the human motion tracking system. First, create and initial the human body models, then we use the Gaussian mixture model to extract the body contour, and use the method of counting pixels belong to human row by row or column by column separately to determine the rectangular location of the body, then extract other features in the rectangular region, and put these features and the human body model as the input of the tracking. Then use our modified particle filter to track. Finally, for the error rate and the time consuming issues in the tracking process, we conducted a comparative experiment.The prototype system was developed on Windows XP platform by Matlab. By comparing the experimental results show that the error rate of our solution in tracking has decreased and the tracking time has reduced.