中文摘要：

针对间断线型或因污迹掩盖、磨损导致的车道标线不连续等情况，在基于前帧图像检测结果设置感兴趣区域的车道标线跟踪算法基础上，通过区域对比度分析并结合自适应大津算法，建立了车道标线有效检测区域的筛选及其自适应阈值图像分割算法。多种典型车道线型下的图像检测试验表明，算法在保持良好实时性的同时，也体现出对不连续车道线型处或破损等因素的鲁棒性和自适应性。

英文摘要：

To solve the problem that sometimes the lane marking is discontinuous because of the broken-line type or smutch or abrasion, a new algorithm based on the method that the series of region of interest （ROI） disposed with the detection results from last frame was raised. The effective detected region was selected with the combination of the regional contrast analysis and OTSU method, and then the image was segmented with adaptive threshold. The experimental result of all kinds of typical lane image detection shows that not only the real time request can be satisfied but also the adaptive capability and the robustness to the influence of discontinuous lane marking can be increased.