Abstract: In order to improve the efficiency of stereo matching and obtain high-precision sub-pixel disparity, a fast stereo matching method based on small baseline is proposed. Firstly, integral image is used to speed up the calculation of adaptive window and normalized cross correlation. Secondly, false match is rejected according to the reliability constraints. Thirdly, sub-pixel matching method base on iterative resample-to-twice scheme is utilized to compute sub-pixel disparity for reliable pixels. Finally, the disparity plane fitting method based on image segmentation is employed to obtain dense sub-pixel disparity map. Experimental results show that the proposed algorithm not only attains high-precision sub-pixel disparity, but also boosts the efficiency of the matching algorithm, and meets the requirements of small baseline based three-dimensional reconstruction.