Oversampled filter banks and predictive subband coders

Authors

Helmut Bölcskei

Reference

PhD thesis, Vienna University of Technology, Nov. 1997.

[BibTeX, LaTeX, and HTML Reference]

Abstract

This thesis is concerned with the theory and applications of oversampled filter banks and predictive subband coders. Filter banks are used in many modern data compression schemes. So far, the interest of the signal processing community has mostly been restricted to critically sampled filter banks. Only recently, there has been an increased interest in oversampled filter banks. We provide a new frame-theoretic treatment of oversampled filter banks and we show that oversampled filter banks have noise reducing properties and yield more design freedom than critically sampled filter banks. We introduce oversampled cosine modulated filter banks, which are practically attractive since they allow an efficient DCT/DST based implementation. Furthermore, a novel image subband coding scheme based on the new class of linear phase even-stacked cosine modulated filter banks is developed. We demonstrate that the proposed subband image coder outperforms existing subband coders based on nonlinear phase cosine modulated filter banks from a perceptual point of view. We provide a detailed noise analysis of oversampled filter banks. Finally, we introduce new highly efficient methods for achieving noise reduction in oversampled filter banks. The resulting oversampled predictive subband coders are attractive for subband coding applications where the resolution of the quantizers used in the subbands is low. In this case our techniques help to drastically improve the effective resolution of the coder.

Keywords

Oversampled filter banks, frame theory, linear operator theory, cosine modulated filter banks, DFT filter banks, Weyl-Heisenberg frames, subband image coding, noise shaping, linear prediction


Download this document:

 

Copyright Notice: © 1997 H. Bölcskei.

This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.