Research on the Mesoscopic Identification Method of Digital Images of Soil-Rock Mixture

摘要

Soil-rock mixture is a mixture of strength weak soil and stiffness large irregular rock block. Identifying accurately soil particle and rock block is one of the main processes of researching on the mesoscopic mechanical properties of the mixture. In this paper, digital image processing technology is used to identify the soil-rock mixture. Firstly, a conceptual model can be built to obtain the numerical statistical law of soil and rock pixel points by the corresponding relationship of pixel points and units in a digital image. Then the digital image can be transformed by cellular automata theory. The result is that soil and rock block pixel points can be identified and island effect can be also removed in the image. In this way, a digital image can be transformed to numerical model effectively.