SAR Image Despeckling by Noisy Reference-Based Deep Learning Method

Ma, Xiaoshuang; Wang, Chen; Yin, Zhixiang; Wu, Penghai

doi:10.1109/tgrs.2020.2990978

Cited by 78 publications

(46 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As temporal stability is rarely observed in practise, areas affected by changes can be masked out in the loss [27] or independent noisy image pairs can be generated from a noisy image thanks to a generative approach [28]. Alternatively, Molini et al [29] present an algorithm enabling direct training on real images, learning to denoise from a single image at a time.…”

Section: Related Workmentioning

confidence: 99%

SAR2SAR: A Semi-Supervised Despeckling Algorithm for SAR Images

Dalsasso

Denis

Tupin

2021

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

Speckle reduction is a key step in many remote sensing applications. By strongly affecting synthetic aperture radar (SAR) images, it makes them difficult to analyse. Due to the difficulty to model the spatial correlation of speckle, a deep learning algorithm with semi-supervision is proposed in this paper: SAR2SAR. Multi-temporal time series are leveraged and the neural network learns to restore SAR images by only looking at noisy acquisitions. To this purpose, the recently proposed noise2noise framework [1] has been employed. The strategy to adapt it to SAR despeckling is presented, based on a compensation of temporal changes and a loss function adapted to the statistics of speckle.A study with synthetic speckle noise is presented to compare the performances of the proposed method with other state-of-theart filters. Then, results on real images are discussed, to show the potential of the proposed algorithm. The code is made available to allow testing and reproducible research in this field.

show abstract

Section: Related Workmentioning

confidence: 99%

SAR2SAR: A Semi-Supervised Despeckling Algorithm for SAR Images

Dalsasso

Denis

Tupin

2021

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

show abstract

“…The development of more powerful computing devices and the increase of data availability has led to substantial advances in machine learning (ML) methods. The use of ML methods allows remote sensing systems to reach high performance in many complex tasks, e.g., despecklization [66][67][68][69][70][71][72][73][74][75][76][77], object detection, semantic segmentation or image classification. These advancements are due to the capability of Deep Neural Networks to automatically learn suitable features from images in a data-driven approach, without manually setting the parameters of specific algorithms.…”

Section: Comparison With Modern Despeckling Methodsmentioning

confidence: 99%

“…The authors of [75] proved that the network is able to learn a clean representation of the image given the noise distributions of the two noisy images are independent and identical. This idea has been employed in SAR despeckling in [76]. The authors make use of multitemporal SAR images of a same area as the input to the noise2noise network.…”

Section: Comparison With Modern Despeckling Methodsmentioning

confidence: 99%

Hyperanalytic Wavelet-Based Robust Edge Detection

2021

View full text Add to dashboard Cite

The imperfections of image acquisition systems produce noise. The majority of edge detectors, including gradient-based edge detectors, are sensitive to noise. To reduce this sensitivity, the first step of some edge detectors’ algorithms, such as the Canny’s edge detector, is the filtering of acquired images with a Gaussian filter. We show experimentally that this filtering is not sufficient in case of strong Additive White Gaussian or multiplicative speckle noise, because the remaining grains of noise produce false edges. The aim of this paper is to improve edge detection robustness against Gaussian and speckle noise by preceding the Canny’s edge detector with a new type of denoising system. We propose a two-stage denoising system acting in the Hyperanalytic Wavelet Transform Domain. The results obtained in applying the proposed edge detection method outperform state-of-the-art edge detection results from the literature.

show abstract

“…Always for the sake of simplicity, this approach is referred to as multitemporal approach in the following of this article. Real data have been also used for training a CNN as in [32] following the Noise2Noise scheme [33]. In such a scheme, the network learns to predict the clean image by using as input-reference data two noisy images with the same underlying clean data but different independent realizations of noise.…”

Section: Introductionmentioning

confidence: 99%

“…Most of these proposals focus only on the definition of the architecture and use very similar cost functions, not taking into account statistical properties of the SAR image and the presence of strong scatterers, demanding their knowledge to the features extraction from the training data. In [25] and [32], the MSE is used as cost function. In [24] and [26], the MSE is combined with a total variation regularization.…”

Section: Introductionmentioning

confidence: 99%

Multi-Objective CNN-Based Algorithm for SAR Despeckling

Vitale

Ferraioli

Pascazio

2021

IEEE Trans. Geosci. Remote Sensing

View full text Add to dashboard Cite

Deep learning (DL) in remote sensing has nowadays become an effective operative tool: it is largely used in applications, such as change detection, image restoration, segmentation, detection, and classification. With reference to the synthetic aperture radar (SAR) domain, the application of DL techniques is not straightforward due to the nontrivial interpretation of SAR images, especially caused by the presence of speckle. Several DL solutions for SAR despeckling have been proposed in the last few years. Most of these solutions focus on the definition of different network architectures with similar cost functions, not involving SAR image properties. In this article, a convolutional neural network (CNN) with a multi-objective cost function taking care of spatial and statistical properties of the SAR image is proposed. This is achieved by the definition of a peculiar loss function obtained by the weighted combination of three different terms. Each of these terms is dedicated mainly to one of the following SAR image characteristics: spatial details, speckle statistical properties, and strong scatterers identification. Their combination allows balancing these effects. Moreover, a specifically designed architecture is proposed to effectively extract distinctive features within the considered framework. Experiments on simulated and real SAR images show the accuracy of the proposed method compared with the state-of-art despeckling algorithms, both from a quantitative and qualitative point of view. The importance of considering such SAR properties in the cost function is crucial for correct noise rejection and details preservation in different underlined scenarios, such as homogeneous, heterogeneous, and extremely heterogeneous.

show abstract

SAR Image Despeckling by Noisy Reference-Based Deep Learning Method

Cited by 78 publications

References 29 publications

SAR2SAR: A Semi-Supervised Despeckling Algorithm for SAR Images

SAR2SAR: A Semi-Supervised Despeckling Algorithm for SAR Images

Hyperanalytic Wavelet-Based Robust Edge Detection

Multi-Objective CNN-Based Algorithm for SAR Despeckling

Contact Info

Product

Resources

About