Kalman particle filters for appearance-based infraed target tracking

An appearance-based infrared target tracking method is proposed under the Kalman particle filter (KPF) framework. In the KPF, the Kalman filter, which can easily incorporate the observation into the state estimation, is used to generate the importance proposal distribution. Therefore, the KPF can be used to robustly track the infrared target against high-speed motion, irregular trajectory, low signal to noise ratio (SNR), and severe sea clutter background. The appearance model, which is constructed by the kernel-based intensity distribution of the infrared target region, is adopted to represent the infrared target. Experimental results and performance comparison show that our proposed method is much effective and robust.