Multi-frames based real-time road detection method for autonomous vehicle

Usually we evaluate a certain road detection method from two aspects, the detecting performance and the executing time. Normally there is a compromise between these two criteria. This paper is focused on how to provide a reliable and real-time road detection result for an autonomous vehicle. By using a vanishing point based high reliability detection method for the leading frame and Hough transformation based real-time method for the following frames, we solve the contradiction of the detecting performance and the executing time. In the vanishing point detection, we use Gabor filters to get the texture information of the road and use particle swarm optimization to increase the efficiency.