Multi-Material Interface Extraction and its Topological Relationship Expression for 3-D Image Data

Abstract

The interface extraction and its topological relationship is a critical step toward the visualization, recognition, and feature retrieval of multi-material objects. The existing methods pay attention only to the interface visualization and extraction, while ignoring the representation of the interface topological relationship. In this article, we proposed a multi-material interface extraction and its topological relationship expression method for 3-D image data. First, we extend the 2-D Canny detector to 3-D for extracting the interface points of 3-D image data. Second, we introduce an efficient clustering method to cluster interface points for obtaining the sub-interfaces. Third, the neighboring relationship of each sub-interface is calculated and the directed skeleton tree with global topological feature is constructed. Finally, we propose a matrix representation to explicitly express the topological relationship of interfaces. Experimental results demonstrate that our method can efficiently extract interfaces for a variety of synthetic and raw scanned 3-D image data, even in the presence of noisy.