Fast and robust L0-tracker using compressive sensing

Abstract

In recent years, Compressive Sensing (CS) or sparse representation has been considered as one of the most favorite topics in the areas of Computer Vision. In particular this theory can be widely applied in Visual Tracking applications. Addressing the problem of sparse representation through minimizations methods can play a dominant role in the CS trackers (trackers based on CS theory). In contrast to the previous algorithms which usually solve the problem of minimization by using L1-norm, L0-norm minimization is used directly to achieve sparseness in our proposed method. Simulations and results demonstrate that the proposed method can achieve the same or better accuracy with many less complexity than traditional algorithms which used interior-point method.