SAR Despeckling Based on CNN and Bayesian Estimator in Complex Wavelet Domain.

Farhadiani, Ramin ORCID: https://orcid.org/0000-0002-3368-7764; Homayouni, Saeid ORCID: https://orcid.org/0000-0002-0214-5356; Bhattacharya, Avik ORCID: https://orcid.org/0000-0001-6720-6108 et Mahdianpari, Masoud ORCID: https://orcid.org/0000-0002-7234-959X (2022). SAR Despeckling Based on CNN and Bayesian Estimator in Complex Wavelet Domain. IEEE Geoscience and Remote Sensing Letters , vol. 19 . p. 4510005. DOI: 10.1109/LGRS.2022.3185557.

Ce document n'est pas hébergé sur EspaceINRS.

Résumé

We propose a hybrid algorithm for despeckling the synthetic aperture radar (SAR) images using the convolutional neural network (CNN) denoising and complex wavelet shrinkage. In particular, we perform the speckle reduction process in the complex wavelet domain. We first despeckled the approximation complex wavelet coefficients using the MUlti-channel LOgarithm with Gaussian denoising (MuLoG) algorithm based on a pretrained CNN model named fast and flexible denoising convolutional neural network (FFDNet). Next, we despeckled the log-transformed details of the complex wavelet coefficients using the averaged version of the maximum a posteriori (AMAP) estimator. The experimental results on simulated and real SAR images showed that the proposed method achieved better speckle suppression in the homogeneous areas while preserving edges and point targets than other state-of-the-art methods.

Type de document:	Article
Mots-clés libres:	convolutional neural network; bayesian estimation; synthetic aperture radar; complex wavelet; synthetic aperture radar despeckling; complex wavelet domain; synthetic aperture radar images; simulated images; maximum a posteriori; homogeneous areas; point target; learning rate; convolutional layers; probability density function; additive noise; digital elevation model; spatial domain; grayscale images; mean ratio; learning-based methods; approximation coefficients; deep learning-based methods; figure of merit; speckle imaging; airborne synthetic aperture radar; contourlet; conv layer; smoothing factor; posterior mode; subband
Centre:	Centre Eau Terre Environnement
Date de dépôt:	20 août 2026 13:39
Dernière modification:	20 août 2026 13:39
URI:	https://espace.inrs.ca/id/eprint/15350

Gestion Actions (Identification requise)

Modifier la notice