Super-resolving material microstructure image via deep learning for microstructure characterization and mechanical behavior analysis

Abstract The digitized format of microstructures, or digital plays a crucial role in modern-day materials research. Unfortunately, the acquisition microstructures through experimental means can be unsuccessful delivering sufficient resolution that is necessary to capture all relevant geometric features microstructures. resolution-sensitive microstructural overlooked due insufficient may limit one’s ability conduct thorough microstructure characterization and material behavior analysis such as mechanical based on numerical modeling. Here, highly efficient super-resolution imaging deep learning developed using residual network super-resolved low-resolution (LR) data for finite element (FE) analysis. Microstructure FE model not only proved accurate those high-resolution (HR) but also provided insights local grain boundary normal stress distribution, which partially considered entirely disregarded LR data-based