Generalized Matrix Spectral Factorization and Quasi-tight Framelets with Minimum Number of Generators

TL;DR

This paper introduces a systematic approach to construct quasi-tight framelets with minimal generators from any compactly supported refinable function, by solving generalized matrix spectral factorization problems for Laurent polynomial matrices.

Contribution

It develops a new theoretical framework linking quasi-tight framelets to generalized matrix spectral factorization, enabling minimal generator construction.

Findings

Constructed quasi-tight framelets with minimal generators

Provided algorithms for spectral factorization and framelet construction

Demonstrated the approach with multiple examples

Abstract

As a generalization of orthonormal wavelets in $L_2(R)$, tight framelets (also called tight wavelet frames) are of importance in wavelet analysis and applied sciences due to their many desirable properties in applications such as image processing and numerical algorithms. Tight framelets are often derived from particular refinable functions satisfying certain stringent conditions. Consequently, a large family of refinable functions cannot be used to construct tight framelets. This motivates us to introduce the notion of a quasi-tight framelet, which is a dual framelet but behaves almost like a tight framelet. It turns out that the study of quasi-tight framelets is intrinsically linked to the problem of the generalized matrix spectral factorization for matrices of Laurent polynomials. In this paper, we provide a systematic investigation on the generalized matrix spectral factorization problem and compactly supported quasi-tight framelets. As an application of our results on generalized matrix spectral factorization for matrices of Laurent polynomials, we prove in this paper that from any arbitrary compactly supported refinable function in $L_2(R)$, we can always construct a compactly supported one-dimensional quasi-tight framelet having the minimum number of generators and the highest possible order of vanishing moments. Our proofs are constructive and supplemented by step-by-step algorithms. Several examples of quasi-tight framelets will be provided to illustrate the theoretical results and algorithms developed in this paper.