Based on Scanning Electron Microscope (SEM), the 2D binary images are obtained and the segmentation algorithm is used to distinguish rock skeleton and pore structure; then, Markov Chain Monte Carlo (MCMC) method is used to reconstruct 3D digital rock, and the corresponding 3D pore network model are extracted, the extraction method of 3D skeleton network model is proposed based on the inverse process and the rock particle size could be analyzed quantitatively; at last, three low permeability rock samples are taken as the example to analyze the structure parameters including particle radius, pore radius and throat radius, and the relationship among the rock structure parameters are described quantitatively. Results show that, the new method for rock particle size analysis based on network model acquires the rock structure parameters quickly, and characterizes the structure parameters quantitatively, which could shorten the data collection time and provide an effective research idea for rock structure parameters analysis.