Obstacle detection by direct estimation of multiple motion ands.cene structure from a moving stereo rig - Systems, Man and Cybernetics, 2003. IEEE International Conference on

In the context ofreal-time collision /obstacle avoidance this paper investigates a direct -method of estimating dense scene structure and image motion. Our proposed method combines both stereo vision (invoIving geometric constraints) and optical flow (involving photometric constraints). It is applied to multiple motion images (i.e., images with vehicle ego-motion plus obstacle motion). Modijications to the direct method were introduced to improve the reliabili$ of the motion estimation. The mult~le-motion-estimation scheme comprises local motion estimations followed by a classification of these estimates in a simplified motion parameter space through cluster analysis. The classification enables segmentation of the dyerent motions that are used to estimate more accurately'.the dense structure and ,motion of objects in a scene. Experimental results on both synthetic and real image sequences demonstrate the potential ofthe method.