Bridge Surface Defect Detection,Screening And Evaluation Method Based On Point Cloud Gaussian Curvature Field

Using high-precision 3D laser scanning technology to evaluate the apparent quality of bridge structures is one of the development directions of bridge detection.At present,the hardware equipment of 3D laser scanners is developing rapidly,but the iterative spe ed of our processing method for the huge laser point cloud data has not adapted to the sp eed of hardware development.For bridge detection,how to use advanced high-precision3 D laser scanners to realize systematic automatic processing of a large amount of point cloud data and rapid detection of bridge surface quality in complex point cloud scenario s,so as to maximize The superiority of the 3D laser scanner is an urgent problem to be s olved at present.In this paper,according to the characteristics of the original point cloud of the brid ge structure and the requirements of the post-processing operation on the bridge point cl oud data,an efficient and accurate point cloud preprocessing algorithm is used to prepro cess the bridge structure point cloud;Research,excavate the Gaussian curvature distribu tion characteristics of the structure appearance,and complete the detection of the appare nt disease of the bridge structure;combined with imaging,use the relevant contour desc ription algorithm to screen out the artificial settings contained in the Gaussian curvature detection results.The main research content of this paper is as follows:First of all,the problems existing in the actual processing of bridge structure point cloud data are analyzed,and the point cloud preprocessing method suitable for bridge st ructure is studied.(1)As the huge original point cloud data of the bridge structure will p ut pressure on the subsequent algorithm processing,the original point cloud image of th e bridge structure is simplified by using the method of curvature down-sampling and bil ateral filtering to denoise;Due to the influence of the registration method,the registratio n results have problems of different efficiency and registration error magnitude.The poi nt cloud fast and accurate registration algorithm fused with the NDT algorithm and the I CP algorithm is used to splicing the multi-site cloud data;(3)Aiming at the shape of the bridge structure The edge will cause a sudden change in the Gaussian curvature feature.The point cloud segmentation method of the fusion of supervoxel and region growing al gorithm is used to segment the point cloud of the bridge structure;(4)By preprocessing t he point cloud data of a test bridge,it is verified that Feasibility of the above point cloud preprocessing methods.Secondly,the characteristics of apparent defects of bridges visualized by point clou ds are analyzed,and the detection method of apparent defects of bridges based on Gauss ian curvature field is studied.(1)In order to restore the apparent real situation of the brid ge structure,the Delaunay triangular mesh algorithm was used to reconstruct the surface of the structural point cloud,and the apparent Gaussian curvature was calculated based on the meshed surface;(2)The bridge deck of the continuous beam test bridge was analy zed Gaussian curvature distribution,in view of the small concave-convex surface forme d by the asphalt concrete on the bridge deck will interfere with the judgment of the disea se result,the method of limiting the interval of the Gaussian curvature calculation result s is used to exclude the small concave-convex surface;Including various forms of disea ses such as crushed stones,pits,and flaking,the diseases are classified by calculating lo cal characteristics such as the number of Gaussian curvature extreme points,the volume density of Gaussian curvature extreme points,and the flatness of Gaussian curvature ex treme points..Then,aiming at the problem that the artificial objects contained in the Gaussian cur vature recognition results will affect the evaluation of the degree of disease,a method fo r screening artificial objects based on the fusion of Hu moments and morphological feat ures was studied.(1)In order to be able to digitally describe the apparent defect shape of the bridge,the joint eigenvector of Hu moment and morphology is used to describe the profile;(2)To extract point clouds of bridge deck diseases,the Gaussian curvature calcul ation results of the bridge deck Carry out binarization processing;(3)In order to screen o ut the artificial installations included in the bridge deck disease,a database of artificial i nstallations is established,and the artificial installations in the bridge deck disease imag e are moved and searched to complete the screening of artificial installations in this area.Finally,taking an arch bridge similar to the spatial point cloud composition and usa ge of the test bridge as the research object,the bridge structure point cloud preprocessin g method,the bridge apparent disease detection method based on Gaussian curvature fie ld and the Hu moment and The screening method of artificial setting objects based on m orphological characteristics is applied to this arch bridge,and the detection of the struct ural apparent disease of the arch bridge is completed.In this paper,the preprocessing method of bridge structure point cloud,the use of Gaussian curvature to detect structural apparent disease,and the completion of manual s etting screening based on the Hu rectangular morphology feature are established.A brid ge appearance based on point cloud Gaussian curvature field The disease detection and screening method provides a new technical path for the 3D laser point cloud technology to efficiently and automatically complete the task of bridge appearance detection.