Visual object tracking using Gaussian process and sparse representation

Abstract

In this paper, we present a new method to solve the object tracking problem in video sequences based on the combination of sparse representation and Gaussian process. Most of sparse representation based trackers only consider the holistic representation and do not make full use of motion information of the target, and hence may fail with more possibility when there is similar object or occlusion in the scene. In this paper we develop a simple yet robust probabilistic tracking model in which the motion information of the target object in the previous frames (this information is captured by Gaussian process) is used to define a prior distribution on the object location in the current frame. Then, by using an appropriate likelihood distribution model (this is done via sparse representation), we can compute the posterior distribution of the object location on the current frame. Experimental results on synthetic and real-world datasets demonstrate the effectiveness of the proposed object tracking algorithm.