基于改进Hough变换的车道线自动识别算法  

Automatic lane line recognition algorithm based on improved Hough transform

作　　者：祖国明 李树壮 李余光 翟双[1] ZU Guoming;LI Shuzhuang;LI Yuguang;ZHAI Shuang(School of Computer Science&Engineering,Changchun University of Technology,Changchun 130102,China)

机构地区：[1]长春工业大学计算机科学与工程学院,吉林长春130102

出　　处：《长春工业大学学报》2024年第2期130-137,共8页Journal of Changchun University of Technology

基　　金：吉林省教育厅基金项目(JJKH20210739KJ)。

摘　　要：针对复杂路况环境中路面图像杂乱或因光照不均导致的车道线不易检测等问题,提出一种基于改进Hough变换的车道线检测算法。首先对采集到的原始图像预处理,利用Canny算子对图像进行边缘增强及区域提取;其次为进一步提高算法的检测时间及准确性,提出利用滑动窗口实现阈值的自适应选取;最后结合渐进概率提出了复杂路况下的车道线检测算法。仿真实验选取KITTI数据集,结果表明,所提算法能够准确提取路面车道线,克服了复杂背景、光照不均带来的影响,与同类算法相比,文中所提的渐进概率Hough变换算法检测时间短、准确率高,具有一定的实用性和优越性。To address the problems of cluttered road images in complex road conditions or lane lines that are not easily detected due to uneven illumination,Proposed an improved Hough transform based lane line detection algorithm.First,the algorithms preprocesses the original image,and uses the Canny operator to perform edge enhancement and region extraction on the image;to further improve the detection time and accuracy of the algorithm,it proposes to use sliding windows to achieve adaptive selection of thresholds;finally,a proposed algorithm for lane line detection under complex road conditions combined with asymptotic probability.The results show that the proposed algorithm can accurately extract the lane lines on the road,overcoming the effects of complex backgrounds and uneven illumination.Compared with similar algorithms,the proposed progressive probability Hough transform algorithm has certain practicality and superiority because of its short detection time and high accuracy.

关 键 词：智能驾驶 车道线检测 边缘检测 HOUGH变换 最小二乘拟合 

分 类 号：TP391.7[自动化与计算机技术—计算机应用技术]