Image-based modeling of granular porous media

We propose a new method of modeling granular media that utilizes a single two- or three-dimensional image and is formulated based on a Markov process. The process is mapped onto one that minimizes the difference between the image and a stochastic realization of the granular medium and utilizes a novel approach to remove possible unphysical discontinuities in the realization. Quantitative comparison between the morphological properties of the realizations and representative examples indicates excellent agreement.