基于路面3D激光点云数据的车辙自动测量与横向定位

Automatic Rutting Measurement and Horizontal Positioning Based on 3D Laser Point Cloud Data of Pavement

针对传统车辙检测方法存在的受车道外杂物、拥包和推波影响,且无法横向定位车辙于车道内位置的问题,提出一种结合路面水平轴和车道边缘线坐标的车辙检测定位方法。研究以2D/3D激光点云数据为研究基础,首先通过改进Canny算法对2D数据中的车道边缘线进行识别定位,从而提取车道内区域数据以消除车道外杂物干扰;然后参照构建的路面水平轴,进行车辙端点和谷底点的搜寻定位;最后结合车辙特征点连续精准测量车辙宽度、最大深度,以及于车道内的横向位置。在实地采集的大量数据基础上,通过与人工测量结果对比得出,该算法具有较高的准确性,其测量的车辙最大深度和宽度与人工检测结果平均相对误差分别为1%和2%,车辙横向定位与人工检测结果相关度高于90%。

The traditional rutting detection method is affected by the debris outside the lane,swells and translation. It is unable to locate the lateral position of the rutting. The improved method proposed in this study combines the horizontal axis of the road and the edge of the lane marking for rutting detection. Based on the 2D/3D laser point cloud data,the improved Canny algorithm is applied to locate the lane edge line using 2D data. The edge points and the valley point of rutting are detected by referencing the constructed road axis. Finally,the width,the maximum depth,and the lateral position of rutting are continuously measured. By comparing with manual measurement,the validation test shows that the proposed method has high accuracy in rutting measurement(the average error of the maximum depth and width measurement are 1% and 2% respectively,and the correlation coefficient for rutting positioning is higher than 90%).