A target mask matrix based moving object tracking method
Yong Zheng Zhou, Yi Li · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2011
The tracking window used by object tracking method based on mean shift algorithm is usually regular in shape. Thus, it is inevitable to introduce some background information into object model. This problem becomes worse while the moving object is irregular in shape. Aiming at this problem, this paper proposes an improved mean shift algorithm based on target mask matrix (MSTMM). TMM is acquired through background subtraction. On the basis, a new method for building object model is proposed. Experiment results show that MSTMM algorithm succeeds in excluding the background information from object model, and the accuracy of moving object localization is improved.