A new efficient vanishing point detection from a single image

Abstract

Detecting the vanishing point from a single road image is a challenging problem since there is only very limited information in the input image which can help the computer to deduce the genuine location of the vanishing point. Moreover, sometimes the cluttered ambient environment in a real road image will hinder rather than help the detection. Learning the advantages and the limitations of current edge-based and texture-based approaches motivates us to propose a new vanishing point detection method, which integrates the efficiency of edge-based methods and the orientation coherence concept of texture-based methods that can be of a great help to improve the accuracy of detection. The proposed method has been tested on over 200 various road images downloaded from the internet. These images exhibit large variations in color, texture, illumination and ambient environment. The experimental results demonstrate that this new method is both efficient and effective at detecting vanishing point compared with some state-of-the-art edge-based and texture-based methods.