deSpeckNet: Generalizing Deep Learning-Based SAR Image Despeckling

Deep learning (DL) has proven to be a suitable approach for despeckling synthetic aperture radar (SAR) images. So far, most DL models are trained to reduce speckle that follows a particular distribution, either using simulated noise or a specific set of real SAR images, limiting the applicability of these methods for real SAR images with unknown noise statistics. In this article, we present a DL method, deSpeckNet,1 that estimates the speckle noise distribution and the despeckled image simultaneously. Since it does not depend on a specific noise model, deSpeckNet generalizes well across SAR acquisitions in a variety of landcover conditions. We evaluated the performance of deSpeckNet on single polarized Sentinel-1 images acquired in Indonesia, The Democratic Republic of Congo, and The Netherlands, a single polarized ALOS-2/PALSAR-2 image acquired in Japan and an Iceye X2 image acquired in Germany. In all cases, deSpeckNet was able to effectively reduce speckle and restore the images in high quality with respect to the state of the art.

[1]  Jong-Sen Lee,et al.  Polarimetric SAR speckle filtering and its implication for classification , 1999, IEEE Trans. Geosci. Remote. Sens..

[2]  Masoud Mahdianpari,et al.  The Effect of PolSAR Image De-speckling on Wetland Classification: Introducing a New Adaptive Method , 2017 .

[3]  Leonard J. Porcello,et al.  Speckle reduction in synthetic-aperture radars , 1976 .

[4]  Henri Maître,et al.  Ratio-Based Multitemporal SAR Images Denoising: RABASAR , 2019, IEEE Transactions on Geoscience and Remote Sensing.

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

[6]  Andrea Vedaldi,et al.  MatConvNet: Convolutional Neural Networks for MATLAB , 2014, ACM Multimedia.

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

[8]  Alfred Stein,et al.  PolSARNet: A Deep Fully Convolutional Network for Polarimetric SAR Image Classification , 2019, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.

[9]  Tara N. Sainath,et al.  Improving deep neural networks for LVCSR using rectified linear units and dropout , 2013, 2013 IEEE International Conference on Acoustics, Speech and Signal Processing.

[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]  Yoshua Bengio,et al.  Understanding the difficulty of training deep feedforward neural networks , 2010, AISTATS.

[12]  Thomas L. Ainsworth,et al.  Improved Sigma Filter for Speckle Filtering of SAR Imagery , 2009, IEEE Transactions on Geoscience and Remote Sensing.

[13]  E. Nezry,et al.  Maximum A Posteriori Speckle Filtering And First Order Texture Models In Sar Images , 1990, 10th Annual International Symposium on Geoscience and Remote Sensing.

[14]  Dong Peng,et al.  A Filter for SAR Image Despeckling Using Pre-Trained Convolutional Neural Network Model , 2019, Remote. Sens..

[15]  Florence Tupin,et al.  NL-SAR: A Unified Nonlocal Framework for Resolution-Preserving (Pol)(In)SAR Denoising , 2015, IEEE Transactions on Geoscience and Remote Sensing.

[16]  Gabriel Vasile,et al.  Intensity-driven adaptive-neighborhood technique for polarimetric and interferometric SAR parameters estimation , 2006, IEEE Transactions on Geoscience and Remote Sensing.

[17]  Laurent Ferro-Famil,et al.  Scattering-model-based speckle filtering of polarimetric SAR data , 2006, IEEE Transactions on Geoscience and Remote Sensing.

[18]  Luisa Verdoliva,et al.  Benchmarking Framework for SAR Despeckling , 2014, IEEE Transactions on Geoscience and Remote Sensing.

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

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

[21]  Alfred Stein,et al.  Scattering property based contextual PolSAR speckle filter , 2017, Int. J. Appl. Earth Obs. Geoinformation.

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

[23]  Giampaolo Ferraioli,et al.  A New Ratio Image Based CNN Algorithm for SAR Despeckling , 2019, IGARSS 2019 - 2019 IEEE International Geoscience and Remote Sensing Symposium.

[24]  Gilles Aubert,et al.  A Variational Approach to Removing Multiplicative Noise , 2008, SIAM J. Appl. Math..

[25]  Jong-Sen Lee,et al.  Intensity and phase statistics of multilook polarimetric and interferometric SAR imagery , 1994, IEEE Trans. Geosci. Remote. Sens..

[26]  Samuel Foucher,et al.  Analysis, Evaluation, and Comparison of Polarimetric SAR Speckle Filtering Techniques , 2014, IEEE Transactions on Image Processing.

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

[28]  Florence Tupin,et al.  MuLoG, or How to Apply Gaussian Denoisers to Multi-Channel SAR Speckle Reduction? , 2017, IEEE Transactions on Image Processing.

[29]  Mihai Datcu,et al.  Evaluation of Bayesian Despeckling and Texture Extraction Methods Based on Gauss–Markov and Auto-Binomial Gibbs Random Fields: Application to TerraSAR-X Data , 2012, IEEE Transactions on Geoscience and Remote Sensing.

[30]  Jong-Sen Lee,et al.  Digital Image Enhancement and Noise Filtering by Use of Local Statistics , 1980, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[31]  LeeJong-Sen.  Digital Image Enhancement and Noise Filtering by Use of Local Statistics , 1980 .

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