A Practical Approach for SAR Image Despeckling Using Deep Learning

In this paper, we propose a deep learning based practical approach to despeckle synthetic aperture radar (SAR) images without using ground truth (despeckled images). Conventionally, the learning based approaches need the despeckled images for the training. However, in real scenarios, it is often difficult to acquire the ground truth. To this end, we demonstrate a technique to restore the speckled images using the available SAR data only. We modify the UNet architecture in order to train it with the given speckled images. We 1) introduce an extra residual connection in each Conv-Block of UNet, and 2) replace the transpose convolution with parameter free bi-linear interpolation. We show the training of this modified network in the mean-squared error (MSE) sense. Experiments have been conducted on ImageNet dataset as well as on the real SAR images. The results are validated both qualitatively as well as quantitatively and compared with state-of-the-art-approaches.

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

[2]  Karen O. Egiazarian,et al.  Image denoising with block-matching and 3D filtering , 2006, Electronic Imaging.

[3]  Alexander A. Sawchuk,et al.  Adaptive Noise Smoothing Filter for Images with Signal-Dependent Noise , 1985, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[4]  Xiao Xiang Zhu,et al.  Deep Learning in Remote Sensing: A Comprehensive Review and List of Resources , 2017, IEEE Geoscience and Remote Sensing Magazine.

[5]  I. Hajnsek,et al.  A tutorial on synthetic aperture radar , 2013, IEEE Geoscience and Remote Sensing Magazine.

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

[7]  Eero P. Simoncelli,et al.  Image quality assessment: from error visibility to structural similarity , 2004, IEEE Transactions on Image Processing.

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

[9]  Thomas Brox,et al.  U-Net: Convolutional Networks for Biomedical Image Segmentation , 2015, MICCAI.

[10]  L. Rudin,et al.  Nonlinear total variation based noise removal algorithms , 1992 .

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

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

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

[14]  Luciano Alparone,et al.  A Tutorial on Speckle Reduction in Synthetic Aperture Radar Images , 2013, IEEE Geoscience and Remote Sensing Magazine.

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