| The Spatial distribution and composition characteristics of rock structure is an important factor to determine the engineering properties of rock mass,and it is also the key control factor to determine the stability and failure mode of rock mass.Traditional measurement method of rock structure cannot get information of rock mass with fast and accurate,which makes the system of rock mass structure information is lack of accurate and complete.It is a method which difficult to support rigorous theory of rock mass structure model in the engineering rock mass of applications.With the development of science and technology,the 3D laser scanning technology in recent years,with its wonderful characteristics such as high precision,long distance and non-contact character,has gradually developed into an important means for the rapid and accurate information collection of rock mass structure.With the help of the terrestrial 3D laser scanning technology(TLS)'s accurate spatial information measurement and computer intelligent recognition vision under digital image technology,we use image data as data source,to identify trace's "skeleton pixels" through the digital image processing technology,combined with point cloud spatial geometric information to get exposed surfaces in the form of trace;we also use point cloud as data source,recognize the structure face through point cloud clustering method.On the basis of comprehensive analysis the theory and knowledge of rock mass structure and rule set construction,the intelligent recognition rule engine is built to realize the automatic/semi-automatic extraction of complex rock mass structure surface information.The main research contents and result of this paper are as follows:(1)Rock trace characteristics extraction based on image processing technology.Using digital image processing technology to get exposed traces.The paper gives a new algorithm which acquaint traces corresponding to the pixel skeleton through CCD image.By using threshold image segmentation,image noise processing,morphological image processing,image thinning,put forward "spread" burr removal method of trace "skeleton pixels".Through trace marking and identification to get the points in the mapping points cloud,finally extract the line feature of rock mass.(2)Rock mass structure extraction based on point cloud and comparative analysis.Exposed structural plane to plane shape for,to high density LIDAR point clouds as the basic data source,this paper presents constraint feature of point cloud based on pumping dilute algorithm,respectively to the normal vector of the law and regional growth method technology,takes the plane form a gel structure surface extraction;research and put forward based on region growth method faces hole repairing method and improve the characteristics of surface integrity;compared two kinds of feature extraction methods,analysis the different threshold difference and the advantages and disadvantages of the.(3)The intelligent recognition of rock mass structural plane based on rule engine.Based on the knowledge of GIS spatial relations,a spatial relation rule engine based on geometric shape is established,and the structure plane of rock mass is correctly identified.According to the clustering algorithm to extract each point cloud cluster,combined with manual intervention to achieve structural plane group and non-structural plane group identification;for digital images to extract the feature trace,trace information in the real world of artificial reasoning by using the rule engine simulations,trace analysis in the space of a geometric shape and spatial relationship(4)Interpretation and statistical analysis of rock mass structure information.According to the structure and trajectory extraction of 3D point cloud collection,digital measuring method of structural plane information.The pole figure,such as dense chart form of occurrence information extraction for statistical analysis of the results,rock mass structural plane and trace data auto/semi-automatic processing and extraction. |
PDF Full Text Request