A Study Of Methods For Road Extraction From Airborne Lidar Data

Road is the critical infrastructure for the development of national economy. It is of vital significant to obtain and update road information timely, accurately and efficiently for the construction of "smart city". With the help of airborne laser radar (LiDAR) technique, high precision of the surface3D point cloud data can be acquired quickly. These data provide an accurate and reliable data source for road information extraction. However, road contour information can’t be extracted accurately using the road extraction methods in existence based on the LiDAR point cloud data, and there are still some problems, including road debris and not high in integrity. To solve these problems and to obtain road information timely and accurately, this paper considers the precision coordinate and echo intensity information obtained from the LiDAR point cloud data, relevant research works are listed as follows:(1) In this paper, characteristics, the data processing procedure of the LiDAR point cloud data is summarized, the principle and main methods of point cloud data filtering is reviewed.(2) Two methods of road extraction are especially detail introduced, the Clode information algorithm and the fuzzy C-means clustering method based on the intensity information.(3) By the comprehensive utilization of the coordinate and echo intensity information of the LiDAR point cloud data, the DBSCAN clustering algorithm is introduced into the constrained TIN method, and based on the mentioned above, the road point cloud information is extracted. And through contrasting analysis with the fuzzy C-means clustering method based on the intensity information, the effectiveness of the proposed method is verified.(4) By the use of the mathematical morphology and Hough transform to process the raster image, which is generated from the extracted road point cloud data, the boundary and center lines of road are extracted, and then the rules of the road network information is obtained. (5) In order to verify the validity and reliability of the methods of road point cloud and road network feature information extraction, the extracting result is statistic analysised by visual interpretation method in the aid of high resolution remote sensing image, and the accuracy, integrity and overall quality is introduced for precision evaluation.It shows that:The method of road extraction based on the3D coordinates and echo intensity information of the LiDAR point cloud data is feasible. The combination of constraint of TIN method and DBSCAN clustering method can remove a lot of roads misclassified points and well keep road point cloud. And on this basis, bye the use of the mathematical morphology and Hough transform to process the raster image of road point cloud data, a complete feature information of road network can be obtained. The research of this paper gives a certain value to roads’information obtaining and updating in the construction of "smart city".