基于级联霍夫变换的车道线快速检测算法

车道线检测是智能辅助驾驶算法中的核心算法之一。为了解决基于传统霍夫变换的车道线检测算法检测效率低下等问题,提出一种基于级联霍夫变换的快速车道线检测算法。该算法首先对视频帧进行ROI选取、滤波、边缘检测、非极大值抑制等预处理,然后使用基于平行坐标系的映射将原始图像转换到参数空间,完成点到线、线到点的映射,接着再使用一次映射,最终实现点到点、线到线的映射,以此快速提取车道线消失点,并根据消失点位置扫描实际车道线,实现车道线的提取。该算法在点的映射过程中,坐标值始终是线性变换,克服了传统霍夫变换在映射过程时需对每一个点进行极坐标转换的缺点,计算更简单,运算效率更高。仿真实验表明,文中提出的改进算法比传统霍夫变换运算速度提高了31%,准确率提高了6.2%,检测效果有明显提高,可广泛应用于智能辅助驾驶中。

A Fast Lane Detection Algorithm Based on Cascade Hough Transform

Lane line detection is one of the core algorithms in the intelligent assisted driving algorithm.In order to solve the problem of low detection efficiency of the traditional lane line detection algorithm based on the Hough transform,we propose a fast lane line detection algorithm based on the cascade Hough transform.The algorithm first performs ROI selection,filtering,edge detection,non-maximum suppression on video frames,and then uses the mapping based on the parallel coordinate system to convert the original image to the parameter space for completion of the point-to-line,line-to-point mapping.And then the mapping is used again to finally achieve point-to-point and line-to-line mapping,so as to quickly extract the vanishing points of the lane line and scan the actual lane line according to the vanishing point position to achieve the lane line extraction.In the point mapping process,the coordinate value is always a linear transformation,which overcomes the shortcomings of traditional Hough transform that requires polar coordinate transformation for each point during the mapping process.The calculation is simpler and the operation efficiency is higher.Simulation experiments show that the improved algorithm proposed is 31%faster than the traditional Hough transform and has an accuracy rate of 6.2%.The detection effect is significantly improved,which can be widely used in intelligent assisted driving.