自适应道路模型的非结构化道路检测算法  

Unstructured road detection algorithm based on adaptive road model

作　　者：许明 张娟[1] 方志军[1] XU Ming;ZHANG Juan;FANG Zhijun(School of Electronic and Electrical Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)

机构地区：[1]上海工程技术大学电子电气工程学院

出　　处：《传感器与微系统》2020年第1期132-135,共4页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(61772328);上海市科委地方院校能力建设项目(15590501300)

摘　　要：为克服非结构化道路中不能准确检测边界线的问题,提出了一种自适应道路模型的非结构化道路检测方法。根据非结构化道路的灰度特征提取感兴趣区域,利用HSV空间图对感兴趣区域使用二维Otsu算法分割,进而提取道路边界点,再针对不同的道路模型自适应选择最小二乘法或随机抽样一致性(RANSAC)算法进行边界点拟合操作。实验结果表明:该方法不仅适用于直线道路,也能检测弯曲道路,且对受阴影影响以及边界不清晰的道路都具有良好的鲁棒性。In order to overcome the problem that the boundaries cannot be accurately detected in unstructured roads,an unstructured road detection method based on adaptive road model is proposed.The algorithm extracts the region of interest(RoI)according to the gray feature of the unstructured road,and uses the HSV space map to segment the region of interest wiTHthe two-dimensional Otsu algorithm,and then the road boundary points can be extracted.Subsequently least square method or random sample consensus(RANSAC)algorithm is used to perform boundary points fitting operation,which is depended on different road models.The experimental results show that the method can be applied not only to straight roads,but also to curved roads,and it has good robustness to roads affected by shadows and road wiTHunclear boundaries.

关 键 词：道路检测 图像分割 最小二乘法 随机抽样一致性(RANSAC)算法 边界拟合 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]