Adaptive phase-retrieval stochastic reconstruction with correlation functions: Three-dimensional images from...

Area of Science:

• Materials Science
• Physics
• Computational Modeling

Background:

• Characterizing 3D heterogeneous media is crucial for understanding structure-property relationships.
• Existing 2D imaging methods lack full 3D structural information needed for modeling.
• There is a need for cost-effective and robust methods to reconstruct 3D structures from 2D data.

Purpose of the Study:

• To develop an adaptive phase-retrieval stochastic reconstruction algorithm (APR) for generating 3D models from 2D images.
• To overcome artifacts present in previous phase-retrieval techniques.
• To enable the hybridization of different reconstruction methods by accommodating various correlation functions and morphological metrics.

Main Methods:

• Developed an adaptive phase-retrieval stochastic reconstruction algorithm (APR).
• Utilized two-point probability and surface-surface functions for optimization during reconstruction.
• Tested APR on three binary porous media samples: sandstone, carbonate, and ceramic.

Main Results:

• APR successfully created 3D replicas from 2D images without characteristic artifacts.
• Reconstruction accuracy for permeability and connectivity was comparable to simulated annealing methods.
• APR demonstrated significant computational efficiency compared to traditional methods.

Conclusions:

• APR provides an accurate and computationally effective method for 3D microstructure reconstruction from 2D data.
• The algorithm's flexibility allows for hybridization with other reconstruction techniques.
• APR can generate rapid, preliminary 3D models for further refinement.