Mean shift tracking with Kernel Co-Occurrence Matrices

We construct Kernel Co-occurrence Matrices (KCMs) to represent the target model and the target candidates. Then those matrices are employed as the tracking cues in mean shift framework. Some improvements are presented in the implementation of the algorithm. First, the angle relation between pixel-pairs is redefined to depict the asymmetric characteristic of the object. Second, the KCMs of the target model and the candidates are normalized to a same integer to increase calculation accuracy. Third, the computation of each pixel weight is modified to improve operation speed. The tracking results of several real world sequences with dark illumination or lighting variance show that the proposed algorithm can track the target effectively.