Kernel-based target tracking with multiple features fusion

Abstract

A novel kernel-based target tracking method with multi-feature fusion is proposed to improve the robustness of target tracking in a complex background. A linear weighted combination of three kernel functions of scale invariant feature transform (SIFT), color and spatial features is applied to represent the probability distribution of the tracked target. SIFT and color features may enhance the target region location stability and accuracy. Meanwhile, the spatial feature is introduced to deal with the target occluded situation. The presented method can handle target scale, orientation, view and illumination changes, and it could also deal with the camera movement mode. Experiments demonstrate that the proposed approach can effectively track the moving target in different scenarios, and could achieve better performance than the classic Camshift algorithm and SIFT tracking approach.