基于道路先验信息和RANSAC算法的车道线检测
|
摘要
为了提高车道线检测算法在光照变化和车道线破损及污迹遮挡等情况下的实时性和鲁棒性,提出了一种基于道路先验信息和随机抽样一致性(RANSAC)算法的检测方法。该算法首先对原始图像进行Sobel边缘检测,提取梯度幅值突变点作为候选车道标志线正负边缘特征点;再基于车道标志线宽度和连续性等先验知识,利用逆透视变换对边缘特征点间的距离进行匹配,去除候选点中的干扰噪声;最后利用RANSAC算法对特征点进行三次曲线拟合。实验结果表明,对各种复杂的城市道路,该识别算法都能准确地识别出车道线。
In order to improve robustness and real-time performance of lane detection algorithm in the conditions of illumination variations,lane breakage and covered by smear,a detection method based on prior knowledge of road and random sample consensus( RANSAC) algorithm is proposed. The algorithm firstly performs the Sobel edge detection on the original image and extracts the points of the gradient amplitude mutation as the feature points of the candidate lane marking line. Based on the prior knowledge of lane marking width and continuity,the inverse perspective mapping( IPM) is used to match the distance between edge feature points to eliminate the interference noise. Afterwards,the feature points were fitted with cubic curve by RANSAC algorithm. The experimental results show that the algorithm can identify the lane markings for all kinds of complicated urban roads accurately.
In order to improve robustness and real-time performance of lane detection algorithm in the conditions of illumination variations,lane breakage and covered by smear,a detection method based on prior knowledge of road and random sample consensus( RANSAC) algorithm is proposed. The algorithm firstly performs the Sobel edge detection on the original image and extracts the points of the gradient amplitude mutation as the feature points of the candidate lane marking line. Based on the prior knowledge of lane marking width and continuity,the inverse perspective mapping( IPM) is used to match the distance between edge feature points to eliminate the interference noise. Afterwards,the feature points were fitted with cubic curve by RANSAC algorithm. The experimental results show that the algorithm can identify the lane markings for all kinds of complicated urban roads accurately.
引文
[1]彭红,肖进胜,沈三明,等.一种基于随机抽样一致性的车道线快速识别算法[J].上海交通大学学报.2014(12):35-40.
[2]WENHUI L,XIAOHU G,YING W.A lane marking detection and tracking algorithm based on sub-regions[C]//Informative and Cybernetics for Computational Social Systems(ICCSS),2014 International Conference on.IEEE,2014:68-73.
[3]樊超,狄帅,侯利龙,等.一种阴影及破损车道线识别方法研究[J].计算机应用研究,2012,29(10):3968-3971.
[4]谢一峰,袁建军,张伟军.车道线实时检测的形态学算法及其DSP实现[J].机电一体化,2013,19(3):26-29.
[5]周涛.单目视觉下车道线检测与跟踪算法的研究与测试[D].华中师范大学,2016.
[6]王超,王欢,赵春霞,等.基于梯度增强和逆透视验证的车道线检测[J].哈尔滨工程大学学报,2014,35(9):1156-1163.
[2]WENHUI L,XIAOHU G,YING W.A lane marking detection and tracking algorithm based on sub-regions[C]//Informative and Cybernetics for Computational Social Systems(ICCSS),2014 International Conference on.IEEE,2014:68-73.
[3]樊超,狄帅,侯利龙,等.一种阴影及破损车道线识别方法研究[J].计算机应用研究,2012,29(10):3968-3971.
[4]谢一峰,袁建军,张伟军.车道线实时检测的形态学算法及其DSP实现[J].机电一体化,2013,19(3):26-29.
[5]周涛.单目视觉下车道线检测与跟踪算法的研究与测试[D].华中师范大学,2016.
[6]王超,王欢,赵春霞,等.基于梯度增强和逆透视验证的车道线检测[J].哈尔滨工程大学学报,2014,35(9):1156-1163.