Vision Assisted Active Safety System for Vehicles on the Highway

Active safety is an important feature of a modern vehicle to provide precautions, warnings or compensatory control before the pre-crash stage of vehicle safety. This paper introduces a vision-based automatic system to detect lane changing and car distance, which use a CCD Camera to capture the vehicle and road image on a PAC DSP board. The algorithm uses multiple image features, such as edges, gradient, vehicle symmetry property, and Hough Transform to detect the Vanishing Point. Some experimental results based on real world video images have been presented, these seem to indicate that the algorithm is capable of identifying a front vehicle and estimating its distance from the host vehicle.