Tracking Distance and Velocity Using a Stereo Vision System

Abstract

In this paper, a method to estimate and track the distance and velocity of an obstacle on the road based on a stereo vision system is presented. The distance and velocity can be calculated using the disparity in the stereo vision system. However, the quantization error of the pixels causes a deterioration in accuracy. The sub-pixel interpolation is used to compensate for the error, and then, the distance and velocity are tracked with a strong tracking extended Kalman filter using a constant velocity model (STEKF-CVM). The Monte-Carlo simulation results show that the performance of STEKF-CVM is better than that of other filters.