Real-time estimation and tracking of optical flow vectors for obstacle detection

Optical flow contains information about the motion of a camera relative to its environment and about the three-dimensional structure of the imaged scene. In this contribution we use that information to detect obstacles in front of a moving vehicle. Since the detection is based on motion no a-priori knowledge about obstacle shape is required. Optical flow vectors are estimated from spatio-temporal derivatives of the gray value function which are computed at video frame rate by the custom-designed hardware MiniVISTA. To eliminate outliers and to speed up obstacle detection by data reduction the estimated vectors are clustered before they are passed to the obstacle test. The purpose of the obstacle test is to separate moving objects from the stationary environment and to separate elevated objects from the ground plane. In continuation of our previous work, obstacle detection is regarded as a state estimation problem. This enables us to enlarge the motion stereo basis by applying a Kalman filter to track optical flow vectors over subsequent image frames. Experimental results obtained from image sequences recorded with our experimental vehicle are presented.