Optimal approximation of multidimensional discrete signals with applications

Abstract

This paper presents an analytical solution to a general approximation problem for multidimensional (m-D) discrete signals. The solution is optimal in the Frobenius-norm sense and includes the well known singular value decomposition (SVD) based approximation as a special case with m=2. Computational issues involved in the algorithm implementation as well as potential applications of the algorithm to image and video coding are discussed.