Moving target-tracking algorithm based on sparse representation and particle filter
Qiu-fen Yang1, Can-jun Li
COMPUTER MODELLING & NEW TECHNOLOGIES 2013 17(5A) 34-40
Department of Computer Science, Hunan Radio &TV University, Changsha, 410004, China
This paper proposes a target tracking algorithm based on 2-dimensional PCA (principal component analysis), which can solve the difficulty of current target tracking algorithm to adapt to the appearance change of target caused by the illumination, shield and position change. First of all, the 2-dimensional PCA method A and sparse representation are used to build the target appearance model, which can reduce the dimension of target; then, by introducing the update method of increment subspace to conduct online update of the target template, it can reduce the algorithm’s requirement of memory space and increase the accuracy of target appearance description; finally, the simulation experiment is conducted. The simulation result shows that compared to other tracking algorithm for moving target, the algorithm proposed in this paper can more accurately track the moving target in the video image, which also shows great robustness to the illumination and position change, and it has significant advantages for the target tracking with serious shield.