|About this Abstract
||2018 TMS Annual Meeting & Exhibition
||2018 Technical Division Student Poster Competition
||SPG-36: Markov Random Field Approach For Three-dimensional Microstructure Reconstruction
||Iman Javaheri, Veera Sundararaghavan
|On-Site Speaker (Planned)
Measurement and analysis of microstructures is an important aspect of material design and structural performance. Grain sizes, cell structures, and precipitate distributions affect the engineering properties as well as the performance of advanced materials. In this work, we present our progress on data-driven methods for microstructure reconstruction using Markov Random Fields. The algorithm reconstructs validated 3D images by matching slices at different voxels with the representative 2D micrographs. This code incorporates an optimization technique that ensures the patches from the 2D micrographs are meshed seamlessly together in the 3D reconstruction. We show that the method effectively models three-dimensional features in the microstructure by using the following three cases: (i) disperse spheres, (ii) anisotropic lamellar microstructure, and (iii) a polycrystalline microstructure. The method is validated by comparing the point probability functions and 3D shape moment invariants of the synthesized images to the original 2D images.