Fast Road Vanishing Point Detection Based on Modified Adaptive Soft Voting

摘要

Detecting the vanishing point from a single image is a challenging task since the information contained in the input image which can be used to detect the location of vanishing point is very limited. In this paper, we propose a framework for vanishing point detection based on the Modified Adaptive Soft Voting (MASV) scheme. First of all, the input image is convolved with the generalized Laplacian of Gaussian (gLoG) filters, which are used to estimate the texture orientation map. Then, MASV scheme is used to get accurate voting map. Finally, peak identification is performed on the voting map to locate the vanishing point. In addition, a scaling method for voting map computation is proposed to further accelerate the vanishing point detection algorithm. Through experiments, we show that the proposed algorithm is about 10 times faster and outperforms by 4.64% on an average than the complete-map based gLoG, which is the state-of-the-art method.