Classification of Volumetric Data Using Multiway Data Analysis

摘要

We introduce a method to extract compressed outline shapes of objects from global textures of volumetric data and to classify them by multiway tensor analysis. For the extraction of outline shapes, we applied three-way tensor principal component analysis to voxel images. A small number of major principal components represent the shape of objects in a voxel image. For the classification of objects, we use tensor subspace method. Using extracted outline shapes and tensor-based classification method, we achieve pattern recognition for volumetric data.