Fragments-based Object Tracking Using Compressive Sensing

摘要

As the Compressive Tracking algorithm is prone to tracking drift and tracking failure under large occlusion, a new algorithm, fragments-based object tracking using compressive sensing is proposed in this paper. Firstly, the candidate region is divided into several fragments. Then different weights are assigned to these fragment classifiers according to their confidence level to eliminate influence of occlusion on tracking result. Finally, a partial fragment classifiers are updated to alleviate the error accumulation caused by occlusion. Experiments show that the algorithm proposed can track object accurately under large occlusion, overcoming the defect of the compressive tracking.