Edge-Weighted Centroid Voronoi Tessellation with Propagation of Consistency Constraint for 3D Grain Segmentation in Microscopic Superalloy Images

Youjie Zhou, Lili Ju, Yu Cao, Jarrell W. Waggoner, Yuewei Lin, Jeff P. Simmons, Song Wang

Abstract: 3D microstructures are important for material scientists to analyze physical properties of materials. While such microstructures are too small to be directly visible to human vision, modern microscopic and serial-sectioning techniques can provide their high-resolution 3D images in the form of a sequence of 2D image slices. In this paper, we propose an algorithm based on the Edge-Weighted Centroid Voronoi Tessellation which uses propagation of the inter-slice consistency constraint. It can segment a 3D superalloy image, slice by slice, to obtain the underlying grain microstructures. With the propagation of the consistency constraint, the proposed method can automatically match grain segments between slices. On each of the 2D image slices, stable structures identified from the previous slice can be well-preserved, with further refinement by clustering the pixels in terms of both intensity and spatial information. We tested the proposed algorithm on a 3D superalloy image consisting of 170 2D slices. Performance is evaluated against manually annotated ground-truth segmentation. The results show that the proposed method outperforms several state-of-the-art 2D, 3D, and propagation-based segmentation methods in terms of both segmentation accuracy and running time.