Exploration Of Automatic Methods For Engineering Rock Mass Classification With Digital Image Processing Technology

Rock mass classification is an effective way to evaluate the quality and stability of engineering rock mass.Most of the multi-factor engineering rock mass classification methods commonly used in China and abroad take parameters related to discontinuities as the most important indicators of rock mass quality evaluation.The characteristics of discontinuities have an important influence and control on the stability and mechanical behavior of the engineering rock mass.The traditional discontinuities measurement method is to manually measure and survey the discontinuities information one by one.This kind of measurement method has the disadvantages of slow information collection speed,high risk and low accuracy and is difficult to meet the future large-scale,automated and high-precision measurement of rock mass discontinuities.In recent years,the rapid development of non-contact measurement methods of rock mass discontinuities such as 3D laser scanning technology,photogrammetry technology and digital image processing technology has provided the possibility of rapid,efficient and accurate discontinuities measurement.However,the data volume of the three-dimensional laser scanning and photogrammetry methods is very large.At present,there are difficulties in the automation and accuracy of the data processing and analysis.Therefore,this paper selects the digital image processing technology with relatively low data volume.We take the image data as the basic research object,and use image processing methods to identify the fracture skeleton of rock mass.Through further analysis of the fracture skeleton,the way to automatically obtain discontinuities parameters in the mainstream engineering rock mass classification is explored.The main research contents and results of this paper are as follows:(1)Rock-fracture skeleton tracing algorithm based on outcrop digital imageAn automatic rock fracture tracing method based on dark region curvilinear structure enhancement of the image is proposed for the characteristics of most rock images with high noise,curvilinear fractures and low contrast at the fractures.This method takes the curvilinear structure enhancement filter Frangi2D as the core in the image preprocessing stage,and improves the contrast of the dark area of the image while reducing noise through various image enhancement methods,greatly enhancing the effect of Frangi2D filtering.In the image segmentation stage,the maximum entropy method is used to binarize the image.Finally,through morphological operations,the final results of the fracture skeleton tracking are obtained.(2)Fracture marking algorithm and extraction and analysis of fracture geometric featuresBased on the obtained fracture skeleton tracking,after removing skeleton burrs,the endpoints,nodes in the skeleton and the coordinates of each point in each fracture are marked in sequence,so as to calculate and analyze each geometric feature of the fractures.(3)Calculation of fractal dimension and JRC of fracture on rock surfaceThe box counting method is used to calculate the fractal dimension of the crack skeleton obtained by image recognition,and the fractal dimension of the single crack and the fractal dimension of the structure surface distribution are calculated respectively.Then the JRC values are obtained from the empirical relationship JRC fractal dimension and rock mass discontinuity roughness.Finally,the basic geometric features and the above parameters are used to try to obtain some engineering rock mass classification results.