BM3D image denoising using heterogeneous computing platforms

Sampsa Sarjanoja; Jani Boutellier; Jari Hannuksela

doi:10.1109/DASIP.2015.7367257

BM3D image denoising using heterogeneous computing platforms

Sampsa Sarjanoja, Jani Boutellier, Jari Hannuksela

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

18 Citations (Scopus)

Abstract

Noise reduction is often performed at an early stage of the image processing path. In order to keep the processing delays small in different computing platforms, it is important that the noise reduction is performed swiftly. In this paper, the block-matching and three-dimensional filtering (BM3D) denoising algorithm is implemented on heterogeneous computing platforms using OpenCL and CUDA frameworks. To our knowledge, these implementations are the first successful open source attempts to use GPU computation for BM3D denoising. The presented GPU implementations are up to 7.5 times faster than their respective CPU implementations. At the same time, the experiments illustrate general design challenges in using massively parallel processing platforms for the calculation of complex imaging algorithms.

Original language	English
Title of host publication	DASIP 2015 - Proceedings of the 2015 Conference on Design and Architectures for Signal and Image Processing
Publisher	IEEE COMPUTER SOCIETY PRESS
Volume	2015-December
ISBN (Electronic)	9791092279108
DOIs	https://doi.org/10.1109/DASIP.2015.7367257
Publication status	Published - 28 Dec 2015
Publication type	A4 Article in conference proceedings
Event	Conference on Design and Architectures for Signal and Image Processing, DASIP 2015 - Cracow, Poland Duration: 23 Sept 2015 → 25 Sept 2015

Conference

Conference	Conference on Design and Architectures for Signal and Image Processing, DASIP 2015
Country/Territory	Poland
City	Cracow
Period	23/09/15 → 25/09/15

Keywords

Image denoising
Mobile computing
Parallel algorithms
Parallel processing

ASJC Scopus subject areas

Computer Graphics and Computer-Aided Design
Computer Vision and Pattern Recognition
Hardware and Architecture
Signal Processing
Electrical and Electronic Engineering

Access to Document

10.1109/DASIP.2015.7367257

Cite this

@inproceedings{4587212fe1d04c74915033449e9ab51c,

title = "BM3D image denoising using heterogeneous computing platforms",

abstract = "Noise reduction is often performed at an early stage of the image processing path. In order to keep the processing delays small in different computing platforms, it is important that the noise reduction is performed swiftly. In this paper, the block-matching and three-dimensional filtering (BM3D) denoising algorithm is implemented on heterogeneous computing platforms using OpenCL and CUDA frameworks. To our knowledge, these implementations are the first successful open source attempts to use GPU computation for BM3D denoising. The presented GPU implementations are up to 7.5 times faster than their respective CPU implementations. At the same time, the experiments illustrate general design challenges in using massively parallel processing platforms for the calculation of complex imaging algorithms.",

keywords = "Image denoising, Mobile computing, Parallel algorithms, Parallel processing",

author = "Sampsa Sarjanoja and Jani Boutellier and Jari Hannuksela",

year = "2015",

month = dec,

day = "28",

doi = "10.1109/DASIP.2015.7367257",

language = "English",

volume = "2015-December",

booktitle = "DASIP 2015 - Proceedings of the 2015 Conference on Design and Architectures for Signal and Image Processing",

publisher = "IEEE COMPUTER SOCIETY PRESS",

note = "Conference on Design and Architectures for Signal and Image Processing, DASIP 2015 ; Conference date: 23-09-2015 Through 25-09-2015",

}

Sarjanoja, S, Boutellier, J & Hannuksela, J 2015, BM3D image denoising using heterogeneous computing platforms. in DASIP 2015 - Proceedings of the 2015 Conference on Design and Architectures for Signal and Image Processing. vol. 2015-December, 7367257, IEEE COMPUTER SOCIETY PRESS, Conference on Design and Architectures for Signal and Image Processing, DASIP 2015, Cracow, Poland, 23/09/15. https://doi.org/10.1109/DASIP.2015.7367257

BM3D image denoising using heterogeneous computing platforms. / Sarjanoja, Sampsa; Boutellier, Jani; Hannuksela, Jari.
DASIP 2015 - Proceedings of the 2015 Conference on Design and Architectures for Signal and Image Processing. Vol. 2015-December IEEE COMPUTER SOCIETY PRESS, 2015. 7367257.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

TY - GEN

T1 - BM3D image denoising using heterogeneous computing platforms

AU - Sarjanoja, Sampsa

AU - Boutellier, Jani

AU - Hannuksela, Jari

PY - 2015/12/28

Y1 - 2015/12/28

N2 - Noise reduction is often performed at an early stage of the image processing path. In order to keep the processing delays small in different computing platforms, it is important that the noise reduction is performed swiftly. In this paper, the block-matching and three-dimensional filtering (BM3D) denoising algorithm is implemented on heterogeneous computing platforms using OpenCL and CUDA frameworks. To our knowledge, these implementations are the first successful open source attempts to use GPU computation for BM3D denoising. The presented GPU implementations are up to 7.5 times faster than their respective CPU implementations. At the same time, the experiments illustrate general design challenges in using massively parallel processing platforms for the calculation of complex imaging algorithms.

AB - Noise reduction is often performed at an early stage of the image processing path. In order to keep the processing delays small in different computing platforms, it is important that the noise reduction is performed swiftly. In this paper, the block-matching and three-dimensional filtering (BM3D) denoising algorithm is implemented on heterogeneous computing platforms using OpenCL and CUDA frameworks. To our knowledge, these implementations are the first successful open source attempts to use GPU computation for BM3D denoising. The presented GPU implementations are up to 7.5 times faster than their respective CPU implementations. At the same time, the experiments illustrate general design challenges in using massively parallel processing platforms for the calculation of complex imaging algorithms.

KW - Image denoising

KW - Mobile computing

KW - Parallel algorithms

KW - Parallel processing

UR - http://www.scopus.com/inward/record.url?scp=84959887479&partnerID=8YFLogxK

U2 - 10.1109/DASIP.2015.7367257

DO - 10.1109/DASIP.2015.7367257

M3 - Conference contribution

AN - SCOPUS:84959887479

VL - 2015-December

BT - DASIP 2015 - Proceedings of the 2015 Conference on Design and Architectures for Signal and Image Processing

PB - IEEE COMPUTER SOCIETY PRESS

T2 - Conference on Design and Architectures for Signal and Image Processing, DASIP 2015

Y2 - 23 September 2015 through 25 September 2015

ER -

BM3D image denoising using heterogeneous computing platforms

Abstract

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this