Obtaining exact depth from binocular disparities is difficult if camera calibration is required. We show that qualitative information can be obtained from stereo disparities with little computation and without prior knowledge (or computation) of camera parameters. First, we derive two expressions that order all matched points in the images by depth in two distinct ways using only image coordinates. Second, we develop an algorithm to compute axes of zero-curvature solely from disparities. -after Author
Bibliographical noteFunding Information:
I thank Tommy Poggio, Jim Little, Davi Geiger, and Walter Gillett for their help. The psychophysical aspect of this work gained from discussions with Terry Boult and Heinrich Bfilthoff. This research was done partly in the MIT AI Laboratory. It was supported by a Chaim Weizmann postdoctoral fellowship, a grant from the Sloan foundation, a grant from the ONR, DARPA, and the NSF.