通过计算双目影像特征的上下视差及其它们的关系，提出了一种对标准松驰算法的相容系数中的各参数的自适应算法。实验结果表明，该算法与松驰法标准计算公式相比大大减少了近代次数，提高了影像匹配的计算速度。

Image matching is the key step of realizing the stereo vision, and the relaxation operating is a powerful way. In some situation, the traditional relaxation operating converges slowly because the process is an iterative procedure. In this paper, authors present an adaptive approach toward relaxation operating and give out the method of determining the correlative coefficient. Finally, the efficiency of algorithm is examined through a practical experiment.