Vision-Based Lane Marking Detection and Moving Vehicle Detection

Abstract

Due to the fast augment of vehicles, traffic accidents become more and more frequent. People's safety is threatened seriously. Vision-based lane marking detection and moving vehicle detection are of great importance to the driving safety. In this paper, we propose a constraint between Sobel operator and Shen Jun edge operator to detect the lane marking points. In order to detect vehicle, we present a new vehicle detect algorithm which uses the shadow under the vehicle and the vertical edge to detect the candidate vehicles. Then SVM+HOG is used to verify them.