基于车道线斜率的车道偏离检测  

Lane Departure Detection Based on the Slope of Lane Lines

作　　者：余厚云[1] 张为公[2] 赵转萍[1] 

机构地区：[1]南京航空航天大学机电学院,南京210016 [2]东南大学仪器科学与工程学院,南京210096

出　　处：《光电工程》2012年第7期43-48,共6页Opto-Electronic Engineering

基　　金：国家教育部博士点新教师基金资助项目(200802861061);江苏省汽车工程重点实验室开放基金资助项目(QC200603);江苏省交通科学研究计划资助项目(06C04)

摘　　要：针对结构化道路上运动车辆的车道偏离检测问题,采用单目视觉作为感知手段,分别从针孔模型下摄像机成像的空间几何关系和图像中车道线消失点的位置特点两个方面出发,推导出车辆直行情况下车道偏离率与两侧车道线斜率比之间的简单函数关系,该函数与摄像机内外参数无关。同时完成了在摄像机不同方向角下的车道偏离率测量试验,结果表明由于行车方向瞬时变化引起的摄像机方向角微小改变对车道偏离检测的影响可以忽略。道路现场试验结果表明,上述视觉测量方法得到的车道偏离率与手工实测结果相比,其相对误差小于5%。基于车道线斜率的单目视觉检测方法避开了繁琐的摄像机参数标定过程,达到了较高的车道偏离检测精度。To resolve the lane departure detection on structural road, monocular vision is adopted as the perception. Respectively based on the verification of projective geometry and the position of the vanishing point of lane lines in the picture with pinhole camera model, the relationship between lane departure and the ratio of lane line＇s slope is concluded as a simple function which is unconcerned with the parameters of camera. Moreover, the experiment of measurement for the quantity of lane departure is performed with different camera orientation angles, and the result shows that the influence of small change of the orientation angles caused by the traffic direction can be neglected. Actual experiment result demonstrates that the precision of lane departure detection by this method is high, and the relative error is less than 5% compared with manual measurement. As stated above, the precision of lane departure detection by the method introduced is high and the complicated calibration of monocular camera can be avoided.

关 键 词：车辆安全 车道偏离 单目视觉 车道线斜率 车道线消失点 

分 类 号：TP212.9[自动化与计算机技术—检测技术与自动化装置]