鲁棒的车辆行道线提取方法

Robust method for extracting of vehicle-tract line

提出了一种新的行道线提取方法。该方法利用均值滤波对道路图像进行亮度估计,把均值滤波后的图像和原图像进行差分从而突出白色的行道线区域,并且采用多阈值方案,对得到的差图像进行二值化。对得到的包含行道线的二值图像进行干扰去除和细化处理,并运用基于加权的Hough变换求得多条候选行道线,基于空间约束从所得的候选行道线中挑选出合适的直线对作为左右行道线。实验结果表明,该算法在复杂路况的情况下能够快速准确地提取出行道线,从而实现对驾车者偏离车道的报警。

A novel street line extraction method is proposed.First,the method uses the average filter to estimate the brightness of the road and then uses the filtered image minus the original image to highlight areas of white-track lines,also uses multi-threshold method to do the subtracted image binaryzation;Second,remove interference and refine the binary image,then use the weighted-Hough transform to obtain a few candidate-track lines.Finally,a pair of the appropriate lines are selected as the left and right-track line by the space constraints.Experimental results show that the method can quickly and accurately extract the travel line from complex road situations,which would warn the drivers when their car deviate the travel line.