Research On Roadway Point Cloud Processing Method Based On 3D Laser Scanning

As one of the most important energy sources in China,coal mine safety has become the top priority.However,coal mine accidents still occur frequently,and accidents caused by roadway surrounding rock deformation are even more difficult to avoid.As an efficient means,3D laser scanning technology has been introduced into the application of coal mine tunnels,such as deformation monitoring,roof bolt classification,etc.,to monitor the stability of roadway and meet the needs of intelligent and less human,which is of great significance for reducing coal mine accidents,ensuring life safety and normal production in coal mines.Through theoretical research,algorithm simulation,experimental verification and analysis,and based on the roadway monitoring based on 3D laser scanning,this thesis has carried out research on the key technology of roadway point cloud processing,and achieved the registration and fusion of roadway point cloud and redundant data filtering,aiming to reduce the occupation of data memory and facilitate the subsequent processing of roadway point cloud.The main research work is as follows:(1)In this thesis,the ranging principle of 3D laser scanning is summarized,and the technical characteristics of 3D laser scanning system are described.In order to achieve the registration and fusion of clouds at different stations,three transformation relationship solving methods,namely singular value decomposition method,orthogonal matrix method and quaternion method,are compared and analyzed.A tunnel point cloud registration scheme is designed to complete rough registration with spherical targets and precise registration with iterative algorithm,so as to achieve better registration effect.(2)In order to eliminate the internal noise caused by dust,water vapor,auxiliary vehicles and personnel flow in the process of laser scanning,two traditional point cloud denoising methods are first elaborated,and then a method for removing the noise point cloud in the roadway is proposed.Finally,simulation comparison and analysis of key parameters are carried out.Under different levels of random noise and obstacle point cloud,this method has good performance,which verifies the feasibility of this method in internal noise removal.(3)In order to eliminate redundant point clouds and reduce the amount of data on the premise of preserving the feature information of the point cloud on the roadway surface,three classical point cloud simplification methods are elaborated first,and then a method of reducing the point cloud on the roadway surface with feature reservation is proposed.Finally,a comparative analysis of an example shows that the simplification effect and precision are better than other methods,which verifies the effectiveness of the proposed method in feature retention and simplification effect.(4)The registration,denoising and reduction methods are tested in real environment or existing data sets.The registration method has achieved high accuracy,and the mean error is 1.4 mm and 1.5 mm respectively.The registration effect is good.The denoising method only takes 10 seconds on the underground tunnel point cloud,reaching 96% accuracy,which is better than the two methods in time and accuracy.The proposed simplification method was applied to both the WHU dataset and the measured tunnel point cloud,and the standard deviation of the simplified model was lower than that of the three classic methods.The effectiveness of the simplification was excellent,verifying the applicability and superiority of the proposed method in this thesis.There are 69 figures,13 tables,and 98 references in this thesis.