基于计算机视觉的车道标线与障碍物自动检测

车道标线与障碍物检测是智能车辆辅助驾驶系统的关键技术问题。论文提出了一种基于计算机视觉的车道和障碍物检测新方法。它根据摄影几何投影变换从图像内容重建出道路平面图,解决了图像中远方车道过于细小、难以检测的缺点,算法对虚线车道特别有效。文中对重建参数选择进行了分析比较,实验表明重建结果对参数变化具有很好的鲁棒性。

Computer Vision-based Automatic Lane and Obstacle Detection

Automatic lane and obstacle detection is one key technique for intelligent vehicle driving.The authors present a novel computer vision-based approach for automatic lane and obstacle detection,which reconstructs the road image that lightens the influence of the small lane segments in the far.This method is very effective,especially for dashed lanes.They also analyze the selection of reconstruction parameters,and prove reconstruction results robust to the parame-ters.