An equivalent pore network model (EPNM) describes complex pore structures in a porous media by statistical parameters. Previous studies using such models have focused on seepage and mechanical dispersion, with few studies considering the effect of molecular diffusion on solute transport. In this study, the convection, molecular diffusion and mechanical dispersion of solutes in porous media were studied using an EPNM to predict the solute transport in porous media. A sensitivity analysis of the model parameters was used to study the effect of the pore structure characteristics on the effective diffusion coefficient of the porous media. The influence of molecular diffusion on the hydrodynamic dispersion was analyzed by comparing numerical results with and without molecular diffusion. The results show that the effective diffusion coefficient, which negatively correlates with the throat curvature and positively correlates with the coordinate number and the connection number ratio, is affected by both the pore volume and the pore-throat diffusion capacity. The molecular diffusion correlates with the convection-induced mechanical dispersion to accelerate the solute transport in the low-velocity region. The results of this study show the microscopic mechanisms influencing molecular diffusion for hydrodynamic dispersion as a theoretical basis for predicting the solute transport flux in pore network models.

With the recent tremendous advances of computer graphics rendering and image editing technologies, computergenerated fake images, which in general do not reflect what happens in the reality, can now easily deceive the inspection of human visual system. In this work, we propose a convolutional neural network (CNN)-based model to distinguish computergenerated (CG) images from natural images (NIs) with channel and pixel correlation. The key component of the proposed CNN architecture is a self-coding module that takes the color images as input to extract the correlation between color channels explicitly. Unlike previous approaches that directly apply CNN to solve this problem, we consider the generality of the network (or subnetwork), i.e., the newly introduced hybrid correlation module can be directly combined with existing CNN models for enhancing the discrimination capacity of original networks. Experimental results demonstrate that the proposed network outperforms state-of-the-art methods in terms of classification performance. We also show that the newly introduced hybrid correlation module can improve the classification accuracy of different CNN architectures.