Robust visual object tracking with extended CAMShift in complex environments

This paper presents a new approach to solve the problem of real-time robust object tracking in complex environments. Generally, traditional CAMShift (Continuous Adaptive Mean Shift) provides speed and robustness for visual tracking, but it will become unstable when similar objects are presented in background or occlusion happens. In this paper, we proposed an advanced two level approach towards these problems with improved back-projection and Kalman filter based occlusion handling. The lower level of the approach implements the multidimensional color histogram and the combination of color and motion information based improved histogram back-projection process. The higher level of the approach implements Kalman filter based occlusion handling process with CAMShift. With this method, our proposed method robustly tracks the target object under rapid illumination changes, highly similar-colored background, and occlusion handling condition in real-time with high accuracy. The experimental results show that the proposed algorithm is robust and efficient to track the various object in indoor/outdoor complex environments.