A novel morphological descriptor for multiphase microstructure reconstruction

The microstructure distribution of multiphase porous media has important effects on its macroscopic properties. In this paper, we propose a framework for multiphase reconstruction based on entropy statistical descriptor, which is used as morphological information to perform on two-dimensional (2D) and three-dimensional (3D) reconstruction. The accuracy of the 2D reconstruction is evaluated using the lineal-path function and two-point cluster function, which reflect the connectivity of the microstructure. From the result of the 2D reconstructions, it is noted that the reconstructed three-phase microstructures have the similar morphological information with that of the original images and their lineal-path function and two-point cluster function are closely similar to each other. For the reconstructed 3D microstructures, the accuracy of the reconstructions is also quantified by the lineal-path function and two-point cluster function, which reflect the comprehensive information of the 3D connectivity of the system. The comparison of the lineal-path function and two-point cluster function between all the slices of the reconstructed 3D structures in three directions and that of the reference images shows that the proposed method can capture the prominent features of the original images. The lineal-path function and two-point cluster function of the reconstructed 3D structures are also compared with that of their original 3D structures, which shows that the reconstructed 3D structures have the similar distribution with that of the original 3D structures. This indicates that our proposed method has the ability to capture salient morphological information of multiphase system.