The Research Of Video Tracking Algorithm

As an important branch of computer vision technology, the video tracking system has got more and more attention from researchers lately. With the development of tracking technology and deeper research about this, video tracking has day by day become a significant part of manufacture field and has a widely applied prospective future.It has already been used widely in many fields such as Human-Computer Interaction, voice recognition, public health, artificial intelligence, and video supervision. Here we mainly investigated an approach of weighted color based particle filter algorithm, and discussed moving objects detection, feature extraction, moving objects tracking and occlusion handling.In terms of moving object detection, we discussed differential method and optical flow algorithm, and then we used frame differential method detecting moving objects.In terms of moving object tracking, we investigated Kalman filter based multi-object tracking algorithm and weighted color histogram based particle filter object tracking algorithm.In the algorithm of Kalman filter, we solved the problems of occlusion and tracking precision by improving the matching function. In the algorithm of particle filter, we solved the Particle degeneracy phenomenon of by integrating resampling technique, and at the same time we used weighted color histogram as object characteristic and Bhattacharyya Distance as comparability function to solve the occlusion problem.