Vanishing Point Detection Under Foggy Weather with Edge-Based Approach

Abstract

Vanishing Point detection is one of the vision-based approaches used for autonomous vehicles and Driver Assistance Systems DAS. It is principally useful for detecting the road needed in vehicle navigation and tracking. Like other methods based on vision, the vanishing point detection approach is deeply sensitive to the presence of bad weather as fog. In this paper, we present an efficient edge-based approach for detecting the vanishing point of road scene under foggy weather based on a combination of an adaptive Canny method for edge detection, and the Hough Transform for straight line extraction. The optimal vanishing point is estimated by applying a k-mean clustering on the candidate points obtained by the straight lines intersection. We tested our approach on 731 real and synthetic images, where the experimental results show that the proposed approach for detecting the vanishing point under foggy weather gives good results.