In order to improve the accuracy of color image stereo matching,an im proved sum of absolute differences (SAD) algorithm is proposed in RGB color sp ace employing Gestalt psychology and color information of pixels.Similarity of color and proximity of distance between pixels in the n eighborhood window are adopted to produce the weight,and linear look-up table method is also used to accelerate the computing speed.The distribution characteristics of RGB components are also considered to produce t he component weights for every pixel,which can further improve the stability of the matching cost functio n.The results of the experiment show that the proposed algorithm improves the precision compared with gray SAD and standard color SAD significantly.