Translation-invariant two-dimensional discrete wavelet transform on graphics processing units

Authors:
Dietmar Wippig;Bernd Klauer
Affiliations:
Ministry of Defence, Rheinbach, Germany and Helmut-Schmidt-University, Hamburg, Germany;Helmut-Schmidt-University, Hamburg, Germany
Venue:
ECS'10/ECCTD'10/ECCOM'10/ECCS'10 Proceedings of the European conference of systems, and European conference of circuits technology and devices, and European conference of communications, and European conference on Computer science
Year:
2010

Citing 9
Cited 0

A Theory for Multiresolution Signal Decomposition: The Wavelet Representation

IEEE Transactions on Pattern Analysis and Machine Intelligence
Characterization of Signals from Multiscale Edges

IEEE Transactions on Pattern Analysis and Machine Intelligence
Accelerating 3D convolution using graphics hardware (case study)

VIS '99 Proceedings of the conference on Visualization '99: celebrating ten years
Exploring the VLSI Scalability of Stream Processors

HPCA '03 Proceedings of the 9th International Symposium on High-Performance Computer Architecture
Stream Processors: Progammability and Efficiency

Queue - DSPs
Scalable Parallel Programming with CUDA

Queue - GPU Computing
NVIDIA Tesla: A Unified Graphics and Computing Architecture

IEEE Micro
Discrete Wavelet Transform on Consumer-Level Graphics Hardware

IEEE Transactions on Multimedia
The JPEG2000 still image coding system: an overview

IEEE Transactions on Consumer Electronics

Quantified Score

Hi-index	0.00

Visualization

Abstract

The Discrete Wavelet Transform (DWT) is used in several signal and image processing applications. Due to the computational expense various approaches have been proposed. One approach is using graphics processing units (GPUs) as stream processors to speed up the calculation of the DWT. This paper presents a GPU implementation of the translation-invariant wavelet transform computed by the "algorithme à trous". Our approach focuses on processing of infrared images, but can be easily used in different image processing applications. We extend our work by the integration of our implementation in wavelet-based edge detection and wavelet denoising. Experiments show that the computation performance could be significantly improved. Initialisation and data transfer are already existing bottlenecks, which could dramatically reduce the GPU performance, if it can't be hided by the application.