Typical Rock Mass Structure Recognition Based On Borehole Optical Image And Radar Image

The complexity of underground rock mass structure brings serious difficulties and challenges to the design and safety of geological engineering such as tunnel construction,bridge design,building quality inspection,underground powerhouse construction,etc.How to accurately detect the structure state of underground rock mass and extract accurate geological structure information is of great practical significance to solve the problems of various geotechnical engineering implementation.Borehole optical image can describe the structure of borehole wall finely,and borehole radar can detect the borehole area in all directions.As borehole imaging technology,the two images have certain similarities.Therefore,it is easy to establish the corresponding relationship between the structural features of the two images,and the fusion of the two data is of great scientific significance to solve the problem of fine detection of underground rock mass structure.This paper combines borehole camera technology and borehole radar technology in engineering geological survey,seeks qualitative description and quantitative representation of typical geological features in borehole optical image and radar image,and studies identification methods of typical rock mass structures such as soil layer,cracks and solution fissures.To realize automatic identification and parameter extraction of typical rock mass structural features,improve the accuracy of borehole photography and ground penetrating radar geological interpretation,and solve the problem of accurate detection and acquisition of deep rock mass structural information urgently needed by deep rock mechanics and engineering.It provides more reliable and accurate analysis methods and means for theoretical analysis and research of rock mechanics behavior,engineering design,construction and analysis,evaluation and prevention of geological hazards.The specific work is as follows:1)By analyzing the color features of typical geological structures in HSV color space in borehole optical images,a method of rock structure recognition based on color features in borehole optical images is proposed to identify soil layer and solution fissure.An adaptive HSV color space detection model is established,which can be used to acquire the depth,area and azimuth of the soil layer or the solution fissure,so as to realize the automatic recognition of the soil layer and the solution fissure in the borehole optical image.2)The gray texture features of various typical rock mass structures in borehole radar images are qualitatively described and quantitatively expressed.For a crack whose reflection feature is an oblique line in radar image,firstly,the radar image is preprocessed.The non-local mean filter with better experimental results is used to remove the Gauss noise.Canny edge detection is used to make the oblique line feature clearer.Then hough transform is improved to detect the oblique line,and then the slope is extracted to obtain the parameters of the structural plane.3)Gray level co-occurrence matrix is used to analyze geological structures,such as karst fissures and soil layers,which are not obvious in borehole radar images.By calculating the energy,contrast,entropy and correlation of the gray level co-occurrence matrix of the image,it is found that the contrast features have the strongest ability to distinguish the geological structure.The contrast and reflection coefficients of the full-aperture radar image are analyzed.It is found that the change of contrast and reflection coefficients at the geological interface is consistent,and can be used as the characteristic parameters to detect the interface of rock mass structure.4)Fusion Analysis of Borehole Optical Image and Radar ImageFirstly,the experimental results of soil layer recognition in borehole optical images using color features are compared with those in radar images using gray texture to verify each other,which proves the accuracy of this method.Secondly,the existing function matching algorithm of sinusoidal curve is improved.Firstly,the optical image of borehole is preprocessed and segmented,and the structural plane is highlighted.By fusing the feature information of borehole radar image,the fitting parameters of sinusoidal function are determined,which improves the accuracy of identification and parameter extraction of rock mass structural plane.5)Combining borehole radar detection data with borehole optical imaging data,a new method for estimating dielectric constant is proposed.According to the relative permittivity determined by the structural plane angle of the optical image,a new definition of "characteristic permittivity" is proposed in this paper.By analyzing the change of characteristic dielectric constant,the structural states of different strata are analyzed.