A method for microstructure similarity clustering and feature reconstruction for weathered weak muddy intercalations

Weak muddy intercalations (WMI) are a type of geo-material with highly unstable mechanical properties, and thus they pose a great threat to the stability of rock slopes and in other rock engineering situations. In this paper, microstructure similarity-based clustering together with image fusion and reconstruction are used to study the microstructures of shallow weathered WMI. The study aims to obtain a reconstructed image of microstructure features that can represent a region to provide the basis for subsequent studies on WMI mechanical properties. The similarity of each microscopic WMI image is calculated using a similarity calculation model based on the microstructure parameters, and images are clustered based on their similarities. Then, image fusion technology is used to combine images in the same cluster. The results are as follows: (1) Similarity corresponding to a cumulative distribution probability of 80% is used as the clustering threshold; (2) The fused WMI microstructure image can represent the microstructure of a layer in the sample. In view of these findings, WMI microstructure clustering and feature reconstruction can provide evidence for studies on WMI lamina structures and failures involving these, which formed the basis for the assessment of the stability of slopes and other situations in which WMI are present.