Avisual Recognition Technology Of Tunnel Excavation Surface Based On Three-dimensional Laser Scanning Point Clouds

Tunnel engineering is a control project for the construction of transportation infrastructure.From design,construction to operation and maintenance,geological structure plays a key role in the safety of the entire life cycle of the tunnel.Especially in the tunnel construction stage,in order to ensure the reasonable construction and the safety of the construction personnel,it is necessary to collect the rock mass information of the tunnel excavation face synchronously to analyze and judge the geological conditions,and adjust the construction measures in time.Therefore,using information technology and professional knowledge to achieve advanced prediction of tunnel geological structure during construction has become a hot technology for tunnel construction.This thesis focuses on the visual recognition technology of tunnel excavation surface based on three-dimensional laser scanning point cloud,which provides technical support for the analysis and identification of rock mass on the tunnel excavation face.Limited by the tunnel construction space and safety requirements,the rock mass information collected on the tunnel excavation face is based on non-contact three-dimensional laser scanning technology.It can not only acquire the rock mass structure point cloud data of the tunnel through long-distance(up to 6000 meters)scan,and overcome the accuracy caused by the discontinuity and anisotropy of the tunnel excavation surface(the measurement accuracy can reach ±2 mm within 10 m).This thesis mainly studies the following key technical problems:1.The isotropic scaling method is used to normalize the 3D laser scanning point cloud data,and the convex hull optimization method of the triangular mesh model after the surface reconstruction of the point cloud data is studied.The isotropic scaling of the massive point cloud data is normalized so that the centroid is at the origin and the average distance between each vertex and the centroid is 1.The Delaunay triangulation reconstruction is used to obtain a triangular mesh model,and its convex hull is optimized to remove sharp slender triangles and ensure that the triangular mesh model is as uniform and smooth as possible.2.Based on the approximate plane of the structural plane,which has a certain scale,the structural plane is identified by using the distance from the scan center point to the triangle surface.Scan to find triangles with close distances,and classify the connected triangles as a structural plane.Finally,the smaller structural surface was removed,and the partial hollow in the structural plane was added to the neighboring structural plane to fill it.3.Combining with multiple program development techniques,C++,Qt,OSG,and CGAL visualization to achieve structural plane recognition.Using Qt 5.8.0 as a program development environment on the Visual Studio 2013 platform,using OSG graphics to realize the threedimensional visualization of point clouds,triangle mesh models,and structural planes,as well as their scaling and rotation operations,realizes the structural plane recognition algorithm.The results of the research show that the structural plane identified by the method of this thesis satisfies the accuracy requirements and achieves the expected results within the allowable error range.