A High-Performance Tree-Block Pipelining Architecture for Separable 2-D Inverse Discrete Wavelet Transform

Yeu-Horng SHIAU; Jer Min JOU

A High-Performance Tree-Block Pipelining Architecture for Separable 2-D Inverse Discrete Wavelet Transform

Yeu-Horng SHIAU, Jer Min JOU

Full Text Views

0

Cite this

Summary :

In this paper, a high-performance pipelining architecture for 2-D inverse discrete wavelet transform (IDWT) is proposed. We use a tree-block pipeline-scheduling scheme to increase computation performance and reduce temporary buffers. The scheme divides the input subbands into several wavelet blocks and processes these blocks one by one, so the size of buffers for storing temporal subbands is greatly reduced. After scheduling the data flow, we fold the computations of all wavelet blocks into the same low-pass and high-pass filters to achieve higher hardware utilization and minimize hardware cost, and pipeline these two filters efficiently to reach higher throughput rate. For the computations of N N-sample 2-D IDWT with filter length of size K, our architecture takes at most (2/3)N² cycles and requires 2N(K-2) registers. In addition, each filter is designed regularly and modularly, so it is easily scalable for different filter lengths and different levels. Because of its small storage, regularity, and high performance, the architecture can be applied to time-critical image decompression.

Publication: IEICE TRANSACTIONS on Information Vol.E86-D No.10 pp.1966-1975

Publication Date: 2003/10/01

Publicized

Online ISSN

DOI

Type of Manuscript: Special Section PAPER (Special Issue on Development of Advanced Computer Systems)

Category

Cite this

Copy

Yeu-Horng SHIAU, Jer Min JOU, "A High-Performance Tree-Block Pipelining Architecture for Separable 2-D Inverse Discrete Wavelet Transform" in IEICE TRANSACTIONS on Information, vol. E86-D, no. 10, pp. 1966-1975, October 2003, doi: .
Abstract: In this paper, a high-performance pipelining architecture for 2-D inverse discrete wavelet transform (IDWT) is proposed. We use a tree-block pipeline-scheduling scheme to increase computation performance and reduce temporary buffers. The scheme divides the input subbands into several wavelet blocks and processes these blocks one by one, so the size of buffers for storing temporal subbands is greatly reduced. After scheduling the data flow, we fold the computations of all wavelet blocks into the same low-pass and high-pass filters to achieve higher hardware utilization and minimize hardware cost, and pipeline these two filters efficiently to reach higher throughput rate. For the computations of N N-sample 2-D IDWT with filter length of size K, our architecture takes at most (2/3)N² cycles and requires 2N(K-2) registers. In addition, each filter is designed regularly and modularly, so it is easily scalable for different filter lengths and different levels. Because of its small storage, regularity, and high performance, the architecture can be applied to time-critical image decompression.
URL: https://global.ieice.org/en_transactions/information/10.1587/e86-d_10_1966/_p

Copy

@ARTICLE{e86-d_10_1966,
author={Yeu-Horng SHIAU, Jer Min JOU, },
journal={IEICE TRANSACTIONS on Information},
title={A High-Performance Tree-Block Pipelining Architecture for Separable 2-D Inverse Discrete Wavelet Transform},
year={2003},
volume={E86-D},
number={10},
pages={1966-1975},
abstract={In this paper, a high-performance pipelining architecture for 2-D inverse discrete wavelet transform (IDWT) is proposed. We use a tree-block pipeline-scheduling scheme to increase computation performance and reduce temporary buffers. The scheme divides the input subbands into several wavelet blocks and processes these blocks one by one, so the size of buffers for storing temporal subbands is greatly reduced. After scheduling the data flow, we fold the computations of all wavelet blocks into the same low-pass and high-pass filters to achieve higher hardware utilization and minimize hardware cost, and pipeline these two filters efficiently to reach higher throughput rate. For the computations of N N-sample 2-D IDWT with filter length of size K, our architecture takes at most (2/3)N² cycles and requires 2N(K-2) registers. In addition, each filter is designed regularly and modularly, so it is easily scalable for different filter lengths and different levels. Because of its small storage, regularity, and high performance, the architecture can be applied to time-critical image decompression.},
keywords={},
doi={},
ISSN={},
month={October},}

Copy

TY - JOUR
TI - A High-Performance Tree-Block Pipelining Architecture for Separable 2-D Inverse Discrete Wavelet Transform
T2 - IEICE TRANSACTIONS on Information
SP - 1966
EP - 1975
AU - Yeu-Horng SHIAU
AU - Jer Min JOU
PY - 2003
DO -
JO - IEICE TRANSACTIONS on Information
SN -
VL - E86-D
IS - 10
JA - IEICE TRANSACTIONS on Information
Y1 - October 2003
AB - In this paper, a high-performance pipelining architecture for 2-D inverse discrete wavelet transform (IDWT) is proposed. We use a tree-block pipeline-scheduling scheme to increase computation performance and reduce temporary buffers. The scheme divides the input subbands into several wavelet blocks and processes these blocks one by one, so the size of buffers for storing temporal subbands is greatly reduced. After scheduling the data flow, we fold the computations of all wavelet blocks into the same low-pass and high-pass filters to achieve higher hardware utilization and minimize hardware cost, and pipeline these two filters efficiently to reach higher throughput rate. For the computations of N N-sample 2-D IDWT with filter length of size K, our architecture takes at most (2/3)N² cycles and requires 2N(K-2) registers. In addition, each filter is designed regularly and modularly, so it is easily scalable for different filter lengths and different levels. Because of its small storage, regularity, and high performance, the architecture can be applied to time-critical image decompression.
ER -