Reliable Extraction Method Of Accurate Rock Discontinuity Based On Terrestrial Lidar Point Clouds

Discontinuity is the basic unit to form the rock mass structure and to control the mechanical properties of rock, thus it is essential to recognize and extract the discontinuity fast in the process of evaluating geological characteristics of rock mass during the process of excavation. Because the discontinuity is irregular and complex, and problems of the conventional extraction are becoming more serious in its reliability, safety, level of automation and great potential to hide the information of the discontinuity surface perpendicular to the rock edge. This paper, by combining the demand and similarity of Surveying and Mapping Science and Engineering Geology, has discussed the limitations and deficiencies of the existing methods in rock mass after the analysis and summary of the current technology and theory in extracting rock mass structure surface. What's more, the theoretical system and process has been established to deal with rock registration, organization and extraction in point clouds, put forward solution to extract rock discontinuity from the ground Lidar point cloud, and finished the design and development of the prototype system in discontinuity extraction, by concerning the geometric characteristics of point cloud data and using the idea of point cloud registration and point cloud segmentation. With two typical cases, the validity, reliability and feasibility of the method has been verified. It has also proved to be feasible to obtain information about the rock discontinuity with the method mentioned in the paper by comparing the precision of occurrence. Moreover, the accuracy can also meet the need of production. The main contents of the research include:(1) After an in-depth research on the natural characteristics of surface of rock mass structure and the analysis of the geometric features of rock discontinuity surface and the characteristic of rock mass point cloud data, the paper points out that small plane characteristics of rock mass is the main feature of rock discontinuity distribution. On this basis, a description is given as to the relationship between the normal vector and rock discontinuity.(2) This essay has classified and discussed the multi-station registration methods. After the comparison and analysis of the characteristics and applicability of three common registration methods, it summarizes the advantages and disadvantages and limitations of the three methods. The whole idea registration method is proposed for rock point cloud data to address the limitations of other methods, namely, the improved particle swarm optimization (PSO)algorithm, using the minimum algebraic sum of normal vector cross products in point cloud data as a fitness function. By the experiment to have an overall registration of the high and steep slope rock point cloud data and comparing the registration results with other classical algorithms. Experiment data is classified into three categories:no noise impact with accurate feature point; having noise impact with inaccurate feature point selection and no noise impact with inaccurate feature point selection. The registration results of the three kinds of data are respectively compared with the results using classical ICP algorithm and ICP algorithm with PSO algorithm's initial registration results. The results of comparative analysis verify the feasibility, the effectiveness and stability of the proposed method.(3) This paper discussed a new principle of rock discontinuity extraction based on point cloud data, and proposed a method for efficient organization of rock point cloud data to maintain the adjacencies primitive between the point and point in terrestrial laser scanning. And this design has achieved a Quadtree-Octree combined index algorithm, which supports efficient browsing and spatial queries.(4) Based on the rock point cloud data organization method, this paper studies the point cloud of the packet-grouped rock surface equation strategy based on the rock discontinuity, judges by the angle between the normal vector of a subset of the criteria,and processes the massive rock point cloud data into packet group, in order to improve the efficiency of the calculation process. this design has achieved a sequential method based on Sequence Normal Vector RANSAC(SEQ-NV-RANSAC) which is an improved detection algorithm to detect whether is compliance with certain constraints between normal vectors point cloud data and calculation of RANSAC plane fitting (NV). This study can improve the accuracy of RANSAC algorithm to extract rock discontinuity.(5) This paper uses the C++programming language, interface library QT, graphics rendering library VTK and point cloud processing third-party libraries PCL (Point Cloud Libary) method herein to achieve the proposed algorithm performed. In order to meet the specific requirements and key steps for rock discontinuity extracted and point cloud data processing, this design has implemented a prototype system of the rock discontinuity surface extracting.(6) Taking the Class2typical rock ground Lidar data as an example, when using the proposed methods and theories to carry out the experimental analysis, we extracted the correct rock discontinuity model, and compared the extracted results with the field of artificial compass measurement results and the foreign scholars research, and we verified the validity,reliability and engineering applicability of the proposed theory and method.