Abstract: A new algorithm integrating SIFT feature points matching into CamShift algorithm was proposed,aiming at tracking object which is prone to failure caused by using general CamShift algorithm under complex backgrounds.The algorithm uses the SIFT feature to realize precise matching of the continuous image sequence which has nothing to do with scale and direction.And it is partially invariant to object scaling,translation and illumination change.Not only it compensates for the lack of taking color as key information of the general CamShift,but also the displacement between the centroid and the center of mass of object tracking window is stable within the threshold.Finally,the effectiveness and stability of the algorithm were verified via comparative experiments.The experimental results show that the new algorithm can achieve stable tracking object against illumination mutations,scaling and rotation under the complex backgrounds.

Key words: Complex backgrounds,CamShift algorithm,SIFT feature points matching,Illumination mutations,Scaling and rotation