摘要
车道线检测时容易受到路面环境的干扰、检测准确度与实时性不易保证。为此,提出了基于稀疏网格和动态特征窗口(DFW)的车道线检测方法。首先在道路区域建立了稀疏网格区域,然后提取了网格上的车道线灰度信息,大幅度排除了冗余像素。利用车道线的方向特性,提出了对称性六向梯度边缘检测方法,进而采用椭圆膨胀元素建立了车道线DFW。利用车道线方向和长度的显著特征,提取了车道线特征边缘并对其进行了Hough直线拟合。在多样性的道路环境中进行了算法测试,讨论了不同分辨率图像的车道线检测耗时。试验表明:提出的算法简单、快速,可以有效排除各类路面干扰像素,能够鲁棒、准识别多种路面环境中的车道线。
Lane detection would be interfered by road environment easily. The accuracy and the real-time performance of lane detection, as a result, is hard to be guaranteed. In this paper a robust lane detection method was proposed based on sparse grid and Dynamic Feature Window（DFW）. The research purpose was to restrain the environment disturbance and to fulfill the real-time requirement. First of all a sparse grid was built up within the road region to extract lane information, and in the meanwhile, exclude the redundant pixels. Next a symmetrical six-direction gradient detection method, which takes advantage of the directional characteristics of lanes, was proposed to obtain DFW. Lane feature edges were then extracted by double threshold constraints of direction and length. Lane feature edges were finally modeled as straight lines by Hough transfornmtior~ The proposed algorithm was tested in various road environments, and moreover, the time consumption of different resolution images was discussed. Experimental results demonstrated that the algorithm could recognize the road lane robustly and rapidly. Most kinds of disturbance from road environment could also be excluded efficiently.
出处
《机械设计与制造》
北大核心
2017年第11期187-190,194,共5页
Machinery Design & Manufacture
基金
陕西省教育厅专项科研计划项目(16JK1375)
西安工业大学校长基金项目(XAGDXJJ15006)
