Abstract: A hierarchical stereo line matching method based on a triangle constraint is proposed to enhance the line matching reliability. The method includes three steps where the first step is to match seed points based on SURF algorithm. The seed points are then used to constrain the nearby line matching. The second step is to triangulate the seed points to reduce the search range of the remaining line matching. A line line matching process is then performed. The third step is a line area matching process under the constraint of the triangles. To reduce the ambiguity of line matching, a shiftable self adaptive line cross correlation method is proposed for discontinuous surfaces and with poor textural information. Experimental studies are carried out with typical close range stereo images and aerial stereo images, and the results reveal that the proposed method is reliable for line matching.