Research On Detection And Reconstruction Of Road Surface Based-on Lynx Vehicle-borne Laser Point Cloud

With the prosperous development of geo-spatial information science and geo-spatial information industry, rapid acquisition and automatic processing of three-dimensional information is becoming the core and frontier of geo-spatial information acquisition technologies. The laser scanning technology which can directly and quickly obtain accurate three-dimensional information is a significant complement of the traditional measurement techniques. Meanwhile, with the gradual civilianization of high-precision positioning and attitude determination products, the development of the mobile mapping system have greatly being accelerated by those new technologies. Shenzhen Institutes of Advanced Technology have undertaken an 863 program: Three-dimensional modeling of large-scale urban scenes based on mobile mapping system. In this thesis, we have done the research on road detection and reconstruction of vehicle-based laser point cloud data as the part of this program.This thesis introduced the Lynx Mobile Mapper system, including the components, working principle and operating processes. We developed the Road Detection and Reconstruction System RoaderDetector in the VS2008 development environment based on OpenGL and Qt Graphic Library using C++ programming language. RoaderDetector realizes the function of reading laser point cloud data, processing input data with normalization and coordinate transformation to conform to the screen displaying purpose, and rendering the processed data in real-time.The implementation methods of main functions in RoaderDetector, road detection and construction, was described detailedly. We extracted the path of scanning car using index file of GPS or the rotation angle of the laser scanner. And then we sampled the extracted path with manually fixed interval, and searched in both direction of each sampled point in the point cloud data, if the elevation of two adjacent points was suddenly changed, it means that the road edge was detected successfully. The abnormalities of the generated road edge were smoothed using two algorithms:the nearest point smoothing and edge points clustering. We can reconstruct the road surface by obtaining the three-dimensional model of road surface with the accurate road edge. The three-dimensional model of road can be exported as 3D object file format to satisfy the demands of accurate road surface in related fields, such as three-dimensional city modeling.The conclutions and innovations of the thesis are as follows:1,We develope the Road Detection and Reconstruction System based on OpenGL and Qt,which enhanced the diffusibility of the system.2,The system realizes the algorithms of road edge detection automatically and road reconstruction.3,The system can solve the abnormalities which occur in road detection partial automatically.4,The three-dimensional model of road can be exported as 3D object file format,which enhance the flexibility of the application and are convenient for the model application user.5,We use laser point-cloud to reconstruct the three-dimensional model of road,which enhance the precision of the model and provide reference for other fields which need high precision.