QUANTITATIVE MICROSTRUCTURE CHARACTERIZATION FOR A RED-BED SOFT ROCK IN THE THREE GORGES RESERVOIR AREA

Red-bed soft rocks are commonly found in the bank slopes of reservoirs in southwestern China. The macroscopic mechanical behavior of red-bed soft rock is closely related to its microscopic particle structure. When studying the physical mechanism of landslide and collapse of red-bed soft rock slope, it is necessary to pay attention to the variation of its microstructure. Both CT scanning and optical microscope can observe the microscopic particle structure of the red-bed soft rock. However, quantitative analysis methods for these microscopic observation images are still rare. In this study, an optical microscope is used to observe a series of sections of the red-bed soft rock in the Three Gorges reservoir area. The image recognition method based on edge detection is adopted to extract the particle structure of the red-bed soft rock section from its microscopic image. A comparative analysis of various edge detection operators reveals that the Canny operator yields the most distinct particle contours with minimal distortion, effectively preserving the original morphology of particles and cement. Furthermore, a non-regular-shaped particle size representation is defined to statistically analyze the particle sizes of red-bed soft rocks. The heterogeneity index calculated using grain size difference was used to quantitatively evaluate the degree of heterogeneity in the microstructure of red-bed soft rock. Comparing the observed images and calculated rock particle data showed that this index accurately reflects the micro-geometric heterogeneity of the microstructure images. These findings provide a method for the quantitative analysis of the microstructure of red-bed soft rocks, offering valuable insights into their mechanical behavior.