摘要
以车载LiDAR数据为对象,提出一种基于多尺度张量投票技术的道路表面裂缝提取方法.首先沿行车路线从剖面图中提取道路路坎,通过行车轨迹线约束提取道路数据.再根据强度和距离信息将道路数据转换成二维特征图像,采用多尺度张量投票法增强特征图像的裂缝信息提取道路表面裂缝.利用点云数据和道路影像数据进行实验验证,结果表明该方法抗噪能力强,裂缝检测质量高.
This paper proposes a multi-scale tensor voting framework that applies tensor voting to mobile laser scanning data to extract pavement cracks. Trajectory data are used to extract road curbs from profiles along the traveUing line to separate road points from non-road points. The extracted road points are interpolated into road feature images. Thus curvilinear cracks are enhanced and extracted with a multi-scale tensor voting framework. Experiments on mobile laser scanning data and road image data were carried out. The results show that the method is robust to noise in both road images and feature images, and can achieve good performance in pavement crack extraction.
出处
《应用科学学报》
CAS
CSCD
北大核心
2015年第5期541-549,共9页
Journal of Applied Sciences
基金
浙江省自然科学基金(No.LQ15D010001)
浙江省教育厅项目基金(No.Y201432349)资助
