Analysis of microscopic structural texture features of cemented sand and gravel layers based on image recognition technology

Abstract To investigate the effects of different surface treatment methods on the microscopic morphology of cemented sand and gravel materials, this study conducts direct shear tests on cemented sand and gravel specimens, with selected samples undergoing SEM-EDS scanning analysis to explore the phase composition and distribution characteristics of the surface structure. The GLCM and Tamura texture methods are used to extract the texture feature parameters of the surface structure under different treatment methods, and these parameters are correlated with the macro-mechanical properties of the materials. A model is established to relate the texture parameters of the surface structure to the mechanical properties of cemented sand and gravel materials. Findings: The surface treatment methods can effectively reduce structural defects, increase the amount of surface gel and calcified materials, and optimize the surface structure. The image texture feature extraction methods allow for effective prediction of the macro-mechanical properties of composite building materials. For the GLCM texture features, the effective parameters for predicting the shear strength of cemented sand and gravel surface structures are energy, entropy, correlation, and contrast. For the Tamura texture features, the effective parameters for predicting shear strength are contrast, directionality, and regularity. The GLCM method serves as a critical indicator for evaluating porosity, defect density, and hydration uniformity. The Tamura method effectively elucidates the texture coarseness and geometric irregularity induced by distinct material constituents. The complementary application of these two methods enables a more comprehensive study of the structural performance of the layer.