基于CHT消失点检测及最小二乘法的车道线检测算法

The Algorithm of Lane Line Detection Based on CHT Vanishing Point Detection and Least Square Method

为解决在车道线识别中检测消失点误差较大和检测速率较慢的不足,提出一种基于CHT消失点检测对车道线采纳点进行最小二乘拟合的车道线检测算法。该算法首先计算RGB熵直方图的最小差值,根据该差值初步分割道路图像。然后引入平行坐标系参数化的级联霍夫变换方法,对查找消失点方法进行改进通过一次投票得出消失点。并以搜寻的消失点为约束条件,利用最小二乘法对最终点进行拟合从而完成对车道线的检测。实验结果表明,所提出的算法对离线图片车道线检测精度达到93.4%,识别速度也有大幅度提升。

For settling the big deviation in detection vanishing point in lane line identification and slow detection speed, this paper proposed an algorithm of lane line detection based on CHT vanishing point detection,which first calculated the least difference value of RGB entropy histogram. The experimental results showed that the accuracy of the proposed algorithm of lane line detection reached 93.4% with significant increase in recognition speed.