Descriptor-based method combined with partition to reconstruct three-dimensional complex microstructures.

Area of Science:

• Materials Science
• Microscopy
• Computational Modeling

Background:

• Reconstructing 3D microstructures from 2D images is crucial for understanding material properties.
• Existing methods struggle with complex microstructures and computational efficiency.

Purpose of the Study:

• To develop an efficient descriptor-based method for 3D microstructure reconstruction from 2D SEM images.
• To apply the method to complex nanoscale pore structures in hardened cement paste.

Main Methods:

• Utilized a descriptor-based method combined with a partition approach and watershed algorithm for feature identification.
• Employed stereological analysis to infer 3D distributions from 2D data.
• Reconstructed 3D microstructures using simulated annealing to match inferred distributions.

Main Results:

• Successfully reconstructed 3D microstructures from 2D SEM images of cement paste.
• Validated the reconstruction accuracy by comparing pore descriptors and two-point correlation functions.
• Demonstrated the ability to quantify material properties, such as absolute permeability, from the 3D model.

Conclusions:

• The proposed method efficiently reconstructs complex 3D microstructures from 2D SEM images.
• The technique accurately captures nanoscale pore structures and enables property quantification.
• This approach significantly improves computational efficiency for microstructure reconstruction.