Real-time tracking method for moving target based on an improved Camshift algorithm

The tracking algorithm of Camshift is only used in simple background with certain color. In this paper, an improved Camshift algorithm is proposed. Firstly the precision localization of the target is realized by wavelet which has the characteristics of multi-resolution analysis and time-frequency local window analysis, then the coordinate of center-of-mass is calculated based on the detected initial frames, and the result is regarded as the initial position of the improved Camshift algorithm. Then the Kalman filter algorithm is combined with Camshift in order to adjust the center-of-mass position of moving target. Experiments results show that the algorithm can effectively overcome interference of the background color, and reliable target tracking in complex situation is realized.