Abstract
Sparse representation has been introduced to visualtracking by finding the best target candidate with mini-mal reconstruction error within the particle filter frame-work. However, most sparse representation based trackershave high computational cost, less than promising trackingperformance, and limited feature representation. To dealwith the above issues, we propose a novel circulant sparsetracker (CST), which exploits circulant target templates.Because of the circulant structure property, CST has the fol-lowing advantages: (1) It can refine and reduce particlesusing circular shifts of target templates. (2) The optimization can be efficiently solved entirely in the Fourier domain. (3) High dimensional features can be embedded into CST to significantly improve tracking performance without sacrificing much computation time. Both qualitative and quantitative evaluations on challenging benchmark sequences demonstrate that CST performs better than all other sparse trackers and favorably against state-of-the-art methods.