Block matching algorithm is useful in many applications such as stereo vision, visual tracking and so on. And the heavy computational burden of the full-search algorithm induces many faster algorithms, which can be classified into two. One class reduces the search area where the other reduces the unnecessary computation of each candidate block. In this paper, a new matching algorithm, which reduces the computational burden by using vote strategy, is proposed. With the observation of the property that the best match also has the similar spatial intensity distribution, block matching algorithm is described. The proposed matching algorithm is fast and robust to speckle noise or object occlusion.