基于道路图像对比度-区域均匀性图分析的自适应阈值算法

针对道路图像检测易受光照及车道标志线曲率等交通环境条件变化干扰而导致检测算法稳定性不佳的问题,在图像阈值分割的经典大津法基础上,对直路、左右转弯、有少量积雪和阴影等多种典型道路行驶图像进行了对比度和区域均匀性图分析,并以此实现了基于大津法的道路图像自适应阈值分割,有效地提高了环境条件变化时道路图像检测算法的鲁棒性。

Adaptive threshold algorithm based on contrast-regional homogeneity analysis of lane image

Aiming at the problem that the road lane line image detection is liable to be interfered by the complicated environmental conditions such as the lighting condition and the lane line curvature, leading to the poor stability of the detection algorithm, images recorded during driving on the various typical road conditions, such as the straight road, the curved road, the lightly snowy road and the shadowed road were analyzed for their contrast and regional homogeneity on the basis of the Otsu image threshold segmentation method. The Otsu adaptive lane image threshold segmentation was implemented based on the classical Otsu segmentation algorithm, improving the robustness of the road lane line image detection under the complicated environmental conditions.