Shape matching algorithm is widely used in many applications of the content-based retrieval

however

due to large data amount of the video stream and the computation of processing

the shape matching algorithm always become bottleneck in the real-time video retrieval. In this paper

we propose an improved algorithm for Hausdorff distance based on multi-resolution

supposed to be applied in the real-time vehicle comparison and recognition

the backward matching algorithm has a great optimization. Experimental resuh shows that the algorithm makes vehicle recognition more fast and accurate

expecially when the model image is big

this improved algorithm has a larger excellence.