3D Reconstruction Of Tunnel Morphology Based On 2D LiDAR

Safe and stable tunnel operation is essential to ensure smooth transportation,and periodic scanning and monitoring of tunnels is a prerequisite to ensure safe and stable tunnel operation.Therefore,it is important to investigate a method that can scan tunnels and generate 3D reconstruction models.Since the traditional 3D laser scanning technology has easily produced data redundancy,the thesis utilized 2D Li DAR to reconstruct the tunnel environment in 3D and improved the traditional algorithm based on the characteristics of the tunnel point cloud data and the problems of the current surface reconstruction method,providing a new method to solve the periodic monitoring of tunnels.The main contents of this thesis are as follows:(1)In order to realize the 3D reconstruction of 2D Li DAR,in this thesis,a 2D Li DAR 3D scanning system was designed.The method of coordinate mapping and transformation between2 D LIDAR and IMU were adopted to map the LIDAR pose to the point cloud corresponding to the 2D plane and realize the 3D of 2D laser point cloud data.The collected data were released to the topic through the ROS system,and the upper computer subscribed to the topic to complete data reception.(2)For the characteristics of laser point cloud data,a pre-processing work was performed on the point cloud data.Due to the large volume of point cloud data,which can seriously affect the speed of subsequent processing,a point cloud streamlining algorithm based on feature retention optimization of the map signal was utilized,and hyperparameters xxx were introduced to control the point cloud density uniformity.The speed and efficiency of the point cloud processing algorithm has been improved while balancing the sharp features of the point cloud data and the homogeneity of the point cloud data to further guarantee the effect of 3D reconstruction visualization.(3)In order to deal with the phenomena of holes and irregular triangles in the traditional surface reconstruction algorithm,the thesis optimized and improved the traditional Poisson reconstruction algorithm by applying the graph signal-based point cloud feature-preserving optimization refinement algorithm and the moving least squares method.The improved algorithm can perform polynomial difference on the streamlined point cloud data by using the moving least squares method to achieve point cloud smoothing and quadratic octree resampling,and then solve Poisson’s equation by further constructing the least equation to complete the high precision reconstruction of 3D point cloud data.Compared with the traditional Poisson reconstruction algorithm and the greedy projection triangulation 3D reconstruction algorithm,the improved MLS Poisson reconstruction algorithm can reconstruct the surface with better flexibility and faster processing speed.(4)Feasibility study of a 2D Li DAR 3D reconstruction system was conducted to collect point cloud data of a highway tunnel,which was transmitted to the host computer and optimized by point cloud data noise reduction,data streamlining and data smoothing.Surface reconstruction experiments were conducted on this data through the improved algorithm of the thesis to realize the tunnel surface reconstruction.The experimental results showed that the 3D model reconstructed for the simulated environment conformed to the real scene,which verified the feasibility of this system for 3D reconstruction in this environment.