由于非结构化道路缺乏明显的车道标识线或车道边界,使得其检测面临很大困难。利用非结构化道路的消失点作为约束,可以显著提高非结构化道路的检测性能。但基于多尺度多方向Gabor纹理的消失点检测方法存在计算量较大的问题,并且背景干扰也常会使得现有的消失点投票方法出现大的误检。为此提出一种基于类Haar特征的非结构化道路纹理快速提取方法,该方法利用设计的实部及虚部矩形模板,通过积分图技术实现了纹理特征的快速复响应;并在此基础上,提出利用正交校正及多样性投票方法,实现非结构化道路消失点的检测。在各种非结构化道路下比较了本文算法与现有两种最新的非结构化道路消失点检测算法,实验结果表明本文算法可以显著提高非结构化道路消失点的检测性能。

The unstructured road detection is a challenging problem due to the fact that many rural or desert roads are not well-paved, lack prominent lane markers or boundaries. By the constraint of vanishing point, the performance of unstructured road detection can be improved significantly. But the existing vanishing point detection methods based on Gabor filters with multiple scales and multiple orientations have high computational cost, and the disturbance of background always deteriorates their performance. In this paper, we propose a new method for detecting the vanishing point of unstructured road based on Haar texture. By using integral image technique, the complex response of Haar texture is fast calculated according to the carefully designed real and imaginary Haar templates, and then the orthogonal rectification and diversity voting methods are used to realize the vanishing point detection. Compared with the state-of-the-art algorithms in a variety of difficult environments, the experimental results demonstrate the effectiveness and efficiency of the proposed method.