RECOVERING 3-DIMENSIONAL VELOCITY AND ESTABLISHING STEREO CORRESPONDENCE FROM BINOCULAR IMAGE FLOWS

Optical flow fields from parallel stereo cameras are analyzed to deter mine the 3-D motion of the camera platform with respect to objects in view and to establish stereo correspondence. A linear system of equati ons, whose coefficients consist of the moments of image velocities and coordinates in the stereo images, is solved to recover the 3-D motion components. This system of equations is derived under the assumption that all feature points have stereo matches (points without stereo mat ches are called drop-ins and drop-outs); however, point-to-point stere o correspondences are not required. No assumptions about the 3-D struc ture of the scene are made. Once the 3-D motion components are compute d, stereo correspondence for a pair of potentially matching points is established by comparing three computed depth values: Z(l), obtained f rom the monocular optical flow equations for the left image; Z(r), obt ained from the same equations for the right image; and Z(s), obtained from stereo disparity. For correct matches these three depths must be nearly equal. The method is applied to numerically generated and labor atory image sequences with and without drop-ins and dropouts, In cases where the percentage of drop-ins and drop-outs is large, an iterative scheme is used to detect and eliminate the drop-ins and drop-outs, th us improving the motion estimates.