Lossless filter banks based on two point transform and interpolative prediction

In this paper, we present a method for designing lossless versions of two-channel FIR filter banks. We demonstrate that equal length PR FIR filter banks can be decomposed into 2-point transforms and unequal length filter banks into interpolative predictions. The lossless versions of the filter banks...

Full description

Saved in:
Bibliographic Details
Main Authors: Komatsu, K., Sezaki, K.
Format: Conference Proceeding
Language:English
Subjects:
Channel bank filters
Design methodology
Discrete cosine transforms
Discrete transforms
Discrete wavelet transforms
Filter bank
Finite impulse response filter
Image coding
Image reconstruction
Performance loss
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this paper, we present a method for designing lossless versions of two-channel FIR filter banks. We demonstrate that equal length PR FIR filter banks can be decomposed into 2-point transforms and unequal length filter banks into interpolative predictions. The lossless versions of the filter banks are obtained by replacing every constituent module by the corresponding lossless version. This method allows construction of the lossless versions of filter banks with arbitrary filter length. Lossless versions of several filter banks are designed and they are found to yield good performance for lossless image compression.
ISSN:1520-6149
2379-190X
DOI:10.1109/ICASSP.1999.756260