New Spectral Decomposition Method for Three-Dimensional Shape Models and Its Applications

Abstract

This paper presents a generalization of a data analysis technique called a singular spectrum analysis (SSA). The original SSA is a tool for analyzing one-dimensional data such as time series, whereas the generalization presented in this paper is suitable for multidimensional data such as three-dimensional polygonal meshes. The basic idea is to generalize the autocorrelation matrix so as to represent mutual relations of multidimensional data flexibly. Two applications of the proposed generalization are also shown briefly.