摘要
为兼顾智能车辆车道识别的鲁棒性和实时性,在摄像机采集的平面车道图像中确定纵横两个方向的二维约束,通过对该二维约束下的局部图像区域进行车道检测,以实现路面车道线的识别。为提高在不同光照条件下车道识别的鲁棒性,提出基于自适应图像阈值的车道识别方法。该方法以车道图像二值化后得到的车道横向宽度为依据,判定之前的二值化阈值是否合适,并经过调整得到适应各种光照的有效阈值。在此基础上进行车道保持系统设计及道路试验。其结果表明,提出的车道识别和跟踪方法能够较好地适应外界光照的变化,其车道识别和跟踪具有较好的实时性和鲁棒性。
In order to obtain both robustness and real by detecting several local areas of the whole image timing for lane recognition of intelligent vehicle, lane recognition is achieved constrained within two dimensions. An adaptive image threshold method is adopted to better recognize the lane under different illumination. It uses the recognized lane width to determine whether the previously-used recognizing threshold is suitable or not, and then the threshold could be adjusted to a suitable value to recognize the path accurately. The lane recognition and tracking system is designed. And the experiments show the intelligent vehicle can recognize and track the lane accurately and robustly under different illumination.
出处
《机械工程学报》
EI
CAS
CSCD
北大核心
2014年第2期146-152,共7页
Journal of Mechanical Engineering
基金
安徽高校省级自然科学研究(KJ2013B074)
安徽科技学院人才引进(稳定)(ZRC2011302)资助项目
关键词
智能车辆
车道保持
二维约束
自适应图像阈值
intelligent vehicle
lane tracking
two-dimension constraints
adaptive image threshold
