Date of Award


Document Type

Open Access Dissertation


Computer Science and Engineering

First Advisor

Wang, Song


Segmentation propagation is the problem of transferring the segmentation of an image to a neighboring image in a sequence. This problem is of particular importance to materials science, where the accurate segmentation of a series of 2D serial-sectioned images of multiple, contiguous 3D structures has important applications. Such structures may have prior-known shape, appearance, and/or topology among the underlying structures which can be considered to improve segmentation accuracy. For example, some materials images may have structures with a specific shape or appearance in each serial section slice, which only changes minimally from slice to slice; and some materials may exhibit specific topology which constrains their structure or neighboring relations.

In this work, we develop a framework for materials image segmentation that segments a variety of materials image types by repeatedly propagating a 2D segmentation from one slice to another, and we formulate each step of this propagation as an optimal labeling problem that can be efficiently solved using the graph-cut algorithm. During this propagation, we propose novel strategies to enforce the shape, appearance, and topology of the segmented structures, as well as handling local topology inconsistency. Most recent works on topology-constrained image segmentation focus on binary segmentation, where the topology is often described by the connectivity of both foreground and background. We develop a new multi-labeling approach to enforce topology in multiple-label image segmentation. In this case, we not only require each segment to be a connected region (intra-segment topology), but also require specific adjacency relations between each pair of segments (inter-segment topology). Finally, we integrate an interactive approach into the proposed framework that improves the segmentation by allowing minimal and simplistic human annotations. We justify the effectiveness of the proposed framework by testing it on various 3D materials images, and we compare its performance against several existing image segmentation methods.