The interpretation of stereo-disparity information: the computation of surface orientation and depth

Two methods for interpreting disparity information are described. Neither requires extraretinal information to scale for distance: one method uses horizontal disparities to solve for the viewing distance, the other uses the vertical disparities. Method 1 requires the assumption that the disparities derive from a locally planar surface. Then from the horizontal disparities measured at four retinal locations the viewing distance and the equation of local surface 'patch' can be obtained. Method 2 does not need this assumption. The vertical disparities are first used to obtain the values of the gaze and viewing distance. These are then used to interpret the horizontal disparity information. An algorithm implementing the methods has been tested and is found to be subject to a perceptual phenomenon known as the 'induced effect'.