In term of underwater the epipolar constraint no longer meet binocular image matching, and the scale-invariant feature transform (SIFT) algorithm can only achieve sparse matching, the underwater stereo matching algorithm based on color segmentation is proposed. The calibration parameters of binocular camera are obtained as well as reference image and image to be matched. Corresponding curve expression of the feature points of reference image on image to be matched is derived and the reference image is segmented by mean shift algorithm. The sum of absolute differences algorithm is used to match two images after assigning different weights to the pixels within the window based on the results of image segmentation and searching for the corresponding point in the curve, which can improve the matching accuracy. Experimental results show that the proposed algorithm is superior to feature matching SIFT algorithm, and the matching accuracy is improved. The area matching algorithm is successfully applied to underwater image dense matching.