Fusion of LIDAR Point Clouds and Large Scale Vector Maps for the Reconstruction of 3-D Road Models

摘要

To describe the environment of a cyber city in detail, 3-D road models are necessary. Due to the diversity of the road types, traditional 2-D vector maps are insufficient to represent 3-D road models when multi-layer road systems are considered. The reconstruction of 3-D road models, thus, becomes an important task in the geoinformatic area. The LIDAR data contains the height information of the road surface. The vector maps record the accurate road boundaries. Thus, we fuse LIDAR point clouds and large scale vector maps to reconstruct the 3-D road models. The proposed scheme comprises two major parts: establishment of 2-D road networks and 3-D road modeling. In the first one, the roadsides of the vector maps are split and merged for the determination of road centerlines and the formation of networks. In the second one, the heights of the road centerlines are derived from LIDAR data to represent the road surface. The profiles of the 3-D roads are obtained by the constraints of slope and slope difference. Then, the road fragments are organized into road models as ribbons. The test data covers Hsin-Chu city in the north of Taiwan. The point density of LIDAR data is 1.78 points/m2. The scale of the vector map is 1:1,000. The experimental results show that the successful rate of the automatic reconstruction for 2-D road network is 81.6%. After the enhancement of the planimetric road networks by manual editing, the reconstructed 3-D road models reach a modeling accuracy better than 0.10m.