Kernel-Diffeomorphism Bayesian Bootstrap Filter to reduce speckle noise on SAR images

Satellite imagery is frequently subject to degradation by noise during both image acquisition and transmission processes. The primary goal of noise reduction techniques is to remove Speckle noise while retaining critical features of the images. In remote sensing applications, Synthetic Aperture Rada...

Full description

Saved in:
Bibliographic Details
Published inComputational statistics Vol. 40; no. 7; pp. 3613 - 3643
Main Authors Zribi, Mourad, Sadok, Ibrahim, Marhaba, Bassel
Format Journal Article
LanguageEnglish
Published Berlin/Heidelberg Springer Berlin Heidelberg 01.09.2025
Springer Nature B.V
Subjects
Bayesian analysis
Economic Theory/Quantitative Economics/Mathematical Methods
Image acquisition
Image quality
Image restoration
Isomorphism
Mathematical models
Mathematics and Statistics
Noise
Noise reduction
Original Paper
Probability and Statistics in Computer Science
Probability Theory and Stochastic Processes
Radar
Radar imaging
Remote sensing
Satellite imagery
Satellites
Signal processing
Statistics
Synthetic aperture radar
Speckle noise
Image restoration
Kernel Diffeomorphism estimator
KDBBF
SAR image
BBF
Online AccessGet full text

Cover

More Information
Summary:Satellite imagery is frequently subject to degradation by noise during both image acquisition and transmission processes. The primary goal of noise reduction techniques is to remove Speckle noise while retaining critical features of the images. In remote sensing applications, Synthetic Aperture Radar (SAR) imagery plays a vital role. Speckle, a granular disturbance typically modelled as multiplicative noise, impacts SAR images as well as all coherent images, resulting in a reduction in image quality. Over the past three decades, numerous techniques have been proposed to mitigate Speckle noise in SAR imagery. This study proposes the Kernel-Diffeomorphism Bayesian Bootstrap Filter (KDBBF) as a novel method for satellite image restoration. The method relies on the multivariate Kernel Diffeomorphism estimator and the Bayesian Bootstrap Filter (BBF). Comparative analyses of the results produced by the new method with those of other image restoration techniques reveal superior performance in Speckle noise reduction in SAR imagery, both quantitatively and qualitatively.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0943-4062
1613-9658
DOI:10.1007/s00180-025-01650-1