Video Tracking Technology Based on Improved Compressed Sensing Algorithm

The video tracking technology is applied widely in military field, intelligent monitoring, security and human-computer interaction fields. In the paper, the neighborhood and dimension of the descriptor of SIFT (Scale Invariant Feature Transform, SIFT) is discussed, then a novel updating strategy of the learning rate of the classifier in the compressive theory is proposed. Target drifting phenomena and occlusion are handled properly after combined with efficient SIFT feature descriptor and improved compressed sensing algorithm. The experiments show that this method not only can improve the real-time performance of tracking target, but also can carry on the tracking of moving target accurately in the event of a target drifting and occlusion.