This paper aims to propose an advanced mean-shift algorithm which based on contour extraction of moving target.Four adjacent frames of images are obtained,the first frame minus the third frame and the second frame minus the fourth frame after registration,and then multiplied by their differences.Morphological processing method was taken to obtain an ideal target area where target color histogram is established after weight calculation.After that,this algorithm calculates the mean-shift vector to track the target.Target color histogram is updated in order to adapt the target deformations when Bhattacharyya coefficient is lower than threshold.The experimental results show that the algorithm with good capability of resisting background interference,and also tracking targets effectively and accurately.This algorithm has strong robustness to both structure irregularity and deformation,it enhance the tracking capability of mean-shift algorithm.