Loughborough University
Leicestershire, UK
LE11 3TU
+44 (0)1509 263171
Loughborough University

Loughborough University Institutional Repository

Please use this identifier to cite or link to this item: https://dspace.lboro.ac.uk/2134/37962

Title: Power shrinkage—curvelet domain image denoising using a new scale-dependent shrinkage function
Authors: Kadri, Oussama
Baarir, Zine-Eddine
Schaefer, Gerald
Keywords: Image noise
Image quality
Image denoising
Curvelet shrinkage
Power shrinkage
Issue Date: 2019
Publisher: © Springer
Citation: KADRI, O., BAARIR, Z-E. and SCHAEFER, G., 2019. Power shrinkage—curvelet domain image denoising using a new scale-dependent shrinkage function. Signal, Image and Video Processing, doi: 10.1007/s11760-019-01484-7
Abstract: © 2019, Springer-Verlag London Ltd., part of Springer Nature. Image processing and analysis algorithms are at the heart of applications in various scientific fields, such as medical diagnosis, military imaging, and astronomy. However, images are typically exposed to noise contamination during their acquisition and transmission. In this paper, we explore recent advancements in image denoising using curvelet domain shrinkage and present a novel scale-dependent shrinkage function, which we call power shrinkage, to enhance restored image quality. Experimental results confirm our proposed method to perform better than classical thresholding and to outperform recent state-of-the-art approaches in denoising different types of noises including speckle, Poisson and additive white Gaussian noise.
Description: This paper is in closed access until 2nd May 2020.
Version: Accepted for publication
DOI: 10.1007/s11760-019-01484-7
URI: https://dspace.lboro.ac.uk/2134/37962
Publisher Link: https://doi.org/10.1007/s11760-019-01484-7
ISSN: 1863-1703
Appears in Collections:Closed Access (Computer Science)

Files associated with this item:

File Description SizeFormat
sivp19.pdfAccepted version10.62 MBAdobe PDFView/Open

 

SFX Query

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.