The virtual time domain depth estimation of stereo-scopic sequence using optical flow

In this paper we present an efficient technique for the stereoscopic video sequence of robot vision, which yields a fast and acceptable solution for the path planning of the robot. To derive the depth information in the two-camera vision system the exploit of the relative geometric discrepancies between two scenes should be made. But finding the corresponding points by using the technique like dynamic programming takes a lot of times. We developed a fast depth estimation algorithm using the optical flow based on the method by Horn and Schunck. The merit of this scheme is that the computational burden can be greatly reduced while obtaining a reasonable solution.