Research On The Body Moving Target Feature Extraction And Tracking Algorithm

Feature extraction and motion tracking in video images is a popular computervision research, and the body moving target feature extraction and motion trackinghas broad application prospects and research value. Now, some researchers havedeveloped many of the practical application of theory and methods. But how to moreaccurately extract human motion characteristics and track movement trends to theinterference of external changes that such as the characteristics of human motion andby the light changes, rotation, and noise is still an unresolved issue. In this paper, weresearch focuses some questions when the frame difference method and the MeanShift algorithm are used to extract and track the human motion.There will be a question when using frame difference method to extract thefeature of human motion: Cavitation will appear when human motion occurs overlapbetween the two frames. So we propose three difference method combined with thehistogram of image enhancement technology to solve these problems. Using thehistogram piecewise linear transformation technique to adjust the video imagesequence in the dynamic range, so that we can highlight the interested part of humangoals from each frame image, and then this part can be extracted by the three framedifference method.Fixed bandwidth affects the tracking results when Mean Shift algorithmtracking the human moving target, and even leads to tracking failure. To solve thisproblem, a pretreatment that transform the image into binary image is taken to abtainthe circular of the human head outline,, and then using Hough transform to detect thesize of its radius.In this paper the bandwidth is adjusted according to the scaling ofthe radius in different frames. The bandwidth can automatically change when usingthe improved Mean Shift algorithm in tracking the human movement away from orclose to the camera and enhance the tracking accuracy.This paper proposes three difference method combined with the histogram ofimage enhancement technology and Mean Shift bases the character of head,and theextensive experiments and analysis are given in every chapter.The experimental results show that the algorithm has strong accuracy and real-time.