Spatio-temporal context tracking algorithm based on master-slave memory space model

Abstract

The spatio-temporal context (STC) tracking algorithm has the advantages of high tracking accuracy and speed, but it may update the target template incorrectly under complex background and interference conditions. A spatio-temporal context tracking algorithm based on master-slave memory space model is proposed in this paper. The algorithm introduces the memory mechanism of Human Visual System (HVS) into the template updating process of STC algorithm, and forms a memory-based update strategy by constructing the master and slave memory spaces. Meanwhile, a method for determining the target location from multi peak points of saliency is proposed. Experimental results indicate that the proposed algorithm has comparatively high accuracy and robustness in the case of the target under occlusion, attitude changes, the target missing and appearing, and illumination changes, etc.