Microstructure reconstruction using diffusion-based generative models

Abstract

This paper proposes a microstructure reconstruction framework with denoising diffusion models for the first time. The novelty and strength of the proposed model lie in its universality and generality for the microstructure characterization and reconstruction (MCR) that can be applied to various types of composite materials. The applicability of the diffusion-based models is validated with several types of microstructures (e.g., polycrystalline alloy, carbonate, ceramics, copolymer, fiber composite, etc.) that have different morphological characteristics. Moreover, an implicit probabilistic model (which yields non-Markovian diffusion processes) is formulated to accelerate the sampling process, thereby controlling the computational cost considering the practicability and reliability.

Keywords:

• microstructure reconstruction
• diffusion model
• denoising diffusion probabilistic model
• neural network
• composite materials

Data availability statement

No data was used for the research described in the article.

Additional information

Funding

This material is based upon work supported by the Air Force Office of Scientific Research under award number FA2386-22-1-4001 and the Institute of Engineering Research at Seoul National University. The authors are grateful for their support.