A real-time object tracking via L2-RLS and compressed Haar-like features matching

In this paper, we present a robust and fast online object tracking algorithm, in which object tracking is achieved by combining L2-regularized least square (L2-RLS) and compressed Haar-like features matching in a Bayesian inference framework. Firstly, the extent of occlusion can be evaluated by L2 tracker. Secondly, the compressed features matching method is used to locate the target object if the extent of occlusion satisfies two inequality constraints. Finally, most of the insignificant samples are removed before computing the compressed features, which makes the computational load of our fused algorithm be only slightly higher than L2 tracker. Both qualitative and quantitative evaluations on numerous challenging image sequences demonstrate that the proposed method is more robust and stable than L2 tracker when the target object undergoes pose variation or rotation, and a paired T-test verifies that it significantly outperforms other state-of-the-art algorithms in terms of accuracy. In addition, our tracker meets the requirement of real-time tracking.