An image-processing-based method for quantitative characterization of freeze-thaw-induced mesoscopic rock structures

Quantitative characterization of rock mesostructures under freeze-thaw cycling remains relatively underdeveloped. This study investigated the quantitative evolution of the meso-structure of Yanshanian granite under freeze-thaw cycles. Scanning electron microscopy (SEM) tests were conducted on rock samples subjected to 0, 10, 20, and 40 freeze-thaw cycles to obtain corresponding SEM images. An image processing method involving image denoising, regional enhancement, convolutional feature extraction, and closing and opening operations was employed to process the images, and the surface crack ratio was statistically analyzed using IPP software. Furthermore, structural complexity analysis of the SEM images was performed by integrating image processing and fractal dimension theory. The proposed image processing method achieved accurate extraction of fracture features in the SEM images. Both the surface crack ratio and fractal dimension showed significant variations with increasing freeze-thaw cycles, indicating the high sensitivity of this method to structural changes in rocks and its suitability for quantifying meso-structural evolution. This study provides a reliable technical approach for comparative analysis of rock meso-structures, particularly applicable for quantitative assessment of structural damage under freeze-thaw cycles.