ABSTRACTPole-like structures (PLSs) located in road environment are important roadway assets. They play a vital role in road safety inspection and road planning. The use of light detection and ranging (lidar) based mobile mapping technology for mapping of PLSs is an important area of research as it holds the potential for automation. Point cloud data of rural, peri-urban, and urban road environment are used in this study, which pose special challenge in view of the complexity of terrain, unlike well-planned roads, which have been the subject of interest in existing literature for identification of PLSs. A new five-step method is proposed in this article. The first two steps, i.e. ground filtering and voxelization of filtered non-ground points, are used for data size reduction. Next three steps are used to extract PLSs from reduced data. The proposed method was tested on point cloud data of three test sites having different levels of complexities. PLSs including partially occluded pole, tilted pole, pole situated very close to other objects, and vertical pole attached to tilted pole were accurately identified. Average correctness and completeness, respectively of 92.6% and 94.9%, were achieved in three different complex test sites, i.e. urban, peri-urban, and rural sites, respectively. Computation complexity shows that our proposed method delivers fast and computationally efficient solution for identifying the PLSs from volumetric mobile lidar point cloud. Impact of PLSs on road safety and road planning is also addressed for these selected test sites.
Read full abstract