Speckle2Void: Deep Self-Supervised SAR Despeckling With Blind-Spot Convolutional Neural Networks

Information extraction from synthetic aperture radar (SAR) images is heavily impaired by speckle noise, hence despeckling is a crucial preliminary step in scene analysis algorithms. The recent success of deep learning envisions a new generation of despeckling techniques that could outperform classical model-based methods. However, current deep learning approaches to despeckling require supervision for training, whereas clean SAR images are impossible to obtain. In the literature, this issue is tackled by resorting to either synthetically speckled optical images, which exhibit different properties with respect to true SAR images, or multi-temporal SAR images, which are difficult to acquire or fuse accurately. In this paper, inspired by recent works on blind-spot denoising networks, we propose a self-supervised Bayesian despeckling method. The proposed method is trained employing only noisy SAR images and can therefore learn features of real SAR images rather than synthetic data. Experiments show that the performance of the proposed approach is very close to the supervised training approach on synthetic data and superior on real data in both quantitative and visual assessments.

[1]  Sergey Ioffe,et al.  Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift , 2015, ICML.

[2]  Xiuhe Li,et al.  SAR image despeckling using a dilated densely connected network , 2018, Remote Sensing Letters.

[3]  Lei Zhang,et al.  Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising , 2016, IEEE Transactions on Image Processing.

[4]  Luisa Verdoliva,et al.  A Nonlocal SAR Image Denoising Algorithm Based on LLMMSE Wavelet Shrinkage , 2012, IEEE Transactions on Geoscience and Remote Sensing.

[5]  E. Nezry,et al.  Structure detection and statistical adaptive speckle filtering in SAR images , 1993 .

[6]  Davide Cozzolino,et al.  Guided Patchwise Nonlocal SAR Despeckling , 2018, IEEE Transactions on Geoscience and Remote Sensing.

[7]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[8]  Jong-Sen Lee,et al.  Speckle analysis and smoothing of synthetic aperture radar images , 1981 .

[9]  Alin Achim,et al.  SAR image denoising via Bayesian wavelet shrinkage based on heavy-tailed modeling , 2003, IEEE Trans. Geosci. Remote. Sens..

[10]  Victor S. Frost,et al.  A Model for Radar Images and Its Application to Adaptive Digital Filtering of Multiplicative Noise , 1982, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[11]  Davide Cozzolino,et al.  Nonlocal CNN SAR Image Despeckling , 2020, Remote. Sens..

[12]  Florian Jug,et al.  Noise2Void - Learning Denoising From Single Noisy Images , 2018, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[13]  Torbjørn Eltoft,et al.  Homomorphic wavelet-based statistical despeckling of SAR images , 2004, IEEE Transactions on Geoscience and Remote Sensing.

[14]  Ye Yuan,et al.  Blind SAR Image Despeckling Using Self-Supervised Dense Dilated Convolutional Neural Network , 2019, ArXiv.

[15]  Jing Zhang,et al.  SAR Image Despeckling Using Multiconnection Network Incorporating Wavelet Features , 2020, IEEE Geoscience and Remote Sensing Letters.

[16]  Jean-Michel Morel,et al.  A Review of Image Denoising Algorithms, with a New One , 2005, Multiscale Model. Simul..

[17]  Vishal M. Patel,et al.  SAR Image Despeckling Using a Convolutional Neural Network , 2017, IEEE Signal Processing Letters.

[18]  Vishal M. Patel,et al.  Generative adversarial network-based restoration of speckled SAR images , 2017, 2017 IEEE 7th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP).

[19]  Dmitri Loguinov,et al.  Bayesian wavelet shrinkage with edge detection for SAR image despeckling , 2004, IEEE Transactions on Geoscience and Remote Sensing.

[20]  Alessandro Foi,et al.  Image Denoising by Sparse 3-D Transform-Domain Collaborative Filtering , 2007, IEEE Transactions on Image Processing.

[21]  Thomas S. Huang,et al.  Non-Local Recurrent Network for Image Restoration , 2018, NeurIPS.

[22]  Jean-Marc Boucher,et al.  Multiscale MAP filtering of SAR images , 2001, IEEE Trans. Image Process..

[23]  Fabrizio Argenti,et al.  Blind Speckle Decorrelation for SAR Image Despeckling , 2014, IEEE Transactions on Geoscience and Remote Sensing.

[24]  Jaakko Lehtinen,et al.  High-Quality Self-Supervised Deep Image Denoising , 2019, NeurIPS.

[25]  Alejandro C. Frery,et al.  Unassisted Quantitative Evaluation of Despeckling Filters , 2017, Remote. Sens..

[26]  Ramesh A. Gopinath,et al.  Wavelet based speckle reduction with application to SAR based ATD/R , 1994, Proceedings of 1st International Conference on Image Processing.

[27]  Ying Li,et al.  HDRANet: Hybrid Dilated Residual Attention Network for SAR Image Despeckling , 2019, Remote. Sens..

[28]  Alexander A. Sawchuk,et al.  Adaptive restoration of images with speckle , 1987, IEEE Trans. Acoust. Speech Signal Process..

[29]  Fawwaz T. Ulaby,et al.  Despeckling SAR images using a low-complexity wavelet denoising process , 2002, IEEE International Geoscience and Remote Sensing Symposium.

[30]  Fawwaz T. Ulaby,et al.  SAR speckle reduction using wavelet denoising and Markov random field modeling , 2002, IEEE Trans. Geosci. Remote. Sens..

[31]  Enrico Magli,et al.  Towards Deep Unsupervised Sar Despeckling with Blind-Spot Convolutional Neural Networks , 2020, IGARSS 2020 - 2020 IEEE International Geoscience and Remote Sensing Symposium.

[32]  Fabrizio Argenti,et al.  Speckle removal from SAR images in the undecimated wavelet domain , 2002, IEEE Trans. Geosci. Remote. Sens..

[33]  Jaakko Lehtinen,et al.  Noise2Noise: Learning Image Restoration without Clean Data , 2018, ICML.

[34]  Loïc Royer,et al.  Noise2Self: Blind Denoising by Self-Supervision , 2019, ICML.

[35]  A. JouanD Speckle Filtering of Sar Images -a Comparative Study between Complex-wavelet-based and Standard Filters , 1997 .

[36]  Robert M. Haralick,et al.  Textural Features for Image Classification , 1973, IEEE Trans. Syst. Man Cybern..

[37]  M. Omair Ahmad,et al.  Spatially Adaptive Wavelet-Based Method Using the Cauchy Prior for Denoising the SAR Images , 2007, IEEE Transactions on Circuits and Systems for Video Technology.

[38]  Fabrizio Argenti,et al.  Multiresolution MAP Despeckling of SAR Images Based on Locally Adaptive Generalized Gaussian pdf Modeling , 2006, IEEE Transactions on Image Processing.

[39]  Jitendra Malik,et al.  A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics , 2001, Proceedings Eighth IEEE International Conference on Computer Vision. ICCV 2001.

[40]  Davide Cozzolino,et al.  SAR image despeckling through convolutional neural networks , 2017, 2017 IEEE International Geoscience and Remote Sensing Symposium (IGARSS).

[41]  Enrico Magli,et al.  DeepSUM: Deep Neural Network for Super-Resolution of Unregistered Multitemporal Images , 2019, IEEE Transactions on Geoscience and Remote Sensing.

[42]  Josiane Zerubia,et al.  SAR image filtering based on the heavy-tailed Rayleigh model , 2006, IEEE Transactions on Image Processing.

[43]  Florence Tupin,et al.  Iterative Weighted Maximum Likelihood Denoising With Probabilistic Patch-Based Weights , 2009, IEEE Transactions on Image Processing.

[44]  Matteo Matteucci,et al.  Deep Learning for SAR Image Despeckling , 2019, Remote. Sens..

[45]  Trevor Darrell,et al.  Fully Convolutional Networks for Semantic Segmentation , 2017, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[46]  Corina da Costa Freitas,et al.  A model for extremely heterogeneous clutter , 1997, IEEE Trans. Geosci. Remote. Sens..

[47]  Xiaoshuang Ma,et al.  Learning a Dilated Residual Network for SAR Image Despeckling , 2017, Remote. Sens..