基于最大相关准则图像分割的结构化道路路径识别和跟踪算法

为提高基于视觉导航的智能车辆对结构化道路车道标识线的识别和跟踪精度,同时消除车流、阴影和光照不均匀等不利因素的影响,提出一种基于最大相关准则的图像分割算法及基于感兴趣区域的车道标识线跟踪算法:首先,对图像进行滤波和光线补偿等前期处理,采用最大相关准则的图像分割算法对道路图像进行阈值分割;然后,根据车道的结构特征及先验知识提取车道标识线的特征点,并运用最小二乘法对特征点拟合,得到车道模型的参数;最后,通过建立感兴趣区域(ROI)的方法实现对车道标识线的准确跟踪。试验结果表明,该算法具有很好的准确性、实时性和鲁棒性。

Structural Road Identification and Tracking Algorithm Based on Maximum Correlation Criterion Image Segmentation

To improve the precision of vision-based structural road lane marker identification and tracking of intelligent vehicle navigation,and eliminate the disadvantages such as vehicle flow,shade and uneven illumination,a method using image segmentation based on Maximum Correlation Criterion(MCC) and lane marker tracking algorithm based on Regions of Interest(ROI) was proposed.First,pre-processing techniques,which include image filtering and ray compensation,were adopted.Then the road images were segmented based on maximum correlation criterion.After that,the parameters of the road model were obtained through fitting the lane marker feature points extracted based on the structure feature and prior knowledge by least-squares method.Finally,the exact lane tracking was implemented accurately by creating ROI.The reliability,real-time property and robustness of the system have been validated.