Research

AGSDNet: Attention and Gradient-Based SAR Denoising Network
In this work, we propose an attention and gradient-based SAR denoising network (AGSDNet) to remove speckle noise from SAR images while preserving finer details. In the proposed network, gradient information of the noisy image is first concatenated with its features in order to increase the feature information content (map). An intermediate feature denoising block (FDB) is then employed to reduce noise from this feature map. Finally, two attention blocks, designed and deployed, in the network focus on preserving the more informative features in the image thereby generating a feature preserved denoised image.

Participating Organisations - IIT Patna India.
Area of Research - Geoscience and Remote Sensing.
Full Text - IEEE Geoscience and Remote Sensing Letters | IEEE Xplore

Gradient and Multi Scale Feature Inspired Deep Blind Gaussian Denoiser
In this work, a novel deep blind Gaussian denoising network is proposed utilizing the concepts of gradient information, multi-scale feature information and feature denoising for removing additive white Gaussian noise (AWGN) from images. The proposed network consists of two modules where in the first module generates an intermediate image whose gradient information is concatenated with the features of second module to generate the final residual image.

Participating Organisations - IIT Patna India.
Area of Research - Image Processing, Computer Vision.
Full Text - IEEE Access | IEEE Xplore

Attention-Based Noise Prior Network for Magnetic Resonance Image Denoising
In this work, we propose a novel approach towards denoising MR images using a combination of attention network and noise level map. The contribution of this work is at three different levels. Firstly Rician noise-level estimation map is fed as prior along with the noisy input data. Secondly, modified U-Net architecture is used to accommodate non-local multi-level and multi-scale features. Thirdly, to preserve long-range dependencies in farther symmetric layers, a symmetric-group attention block is used.

Participating Organisations - IIT Patna India.
Area of Research - Medical Imaging.
Full Text - IEEE International Symposium on Biomedical Imaging ISBI 2022 | IEEE Xplore

Bayesian Approach in a Learning-Based Hyperspectral Image Denoising Framework
In this work, we design the training paradigm emphasizing the role of loss functions in neural network; similar to as observed in model-based optimization methods. Further, Bayesian motivated loss functions also act as priors to constrain the solution space to the types of noise observed in hyperspectral image acquisition process. As a result, loss functions derived in Bayesian setting and employed in neural network training boosts the denoising performance.

Participating Organisations - IIT Patna India, INRIA Bordeaux France.
Area of Research - Geoscience and Remote Sensing.
Full Text - IEEE Access | IEEE Xplore  |  HAL INRIA  |  Bibtex

Feature based Denoising for Fluorescence Microscopy
In this work, we propose an image denoising algorithm based on the concept of feature extraction through multifractal decomposition and then estimate a noise free image from the gradients restricted to these features. The results obtained have surpassed in quality over the best algorithms currently available.

Participating Organisations - IIT Patna India, INRIA Bordeaux France.
Area of Research - Medical Imaging.
Full Text - Scientific Reports | Nature.com  |  HAL INRIA  |  Bibtex

Photo-realistic Reconstruction of Camera Images
Camera images are generally limited by the diffraction limit of light, due to which they resultant images are blurred and noisy. In this work we propose two different techniques for reconstruction of these noisy images using an optimization based primal dual framework.

Participating Organisations - IIT Patna India, MRC Lab of Molecular Biology Cambridge, UK .
Area of Research - Image Processing, Computer Vision.

Denoised High Resolution Generative Adversarial Network - DHRGAN
The advent of deeper convolutional neural networks and related methodologies have made significant achievements in the area of single image super-resolution (SISR). However, none of these techniques are equipped to handle noisy images. In this work, we propose a denoised high resolution generative adversarial network (DHRGAN), capable of handling noise removal from given sample images while trying to super-resolve it to the desired magnification.

Participating Organisations - IIT Patna India.
Area of Research - Machine Learning.
Full Text - IEEE Xplore  |  Bibtex

Structure Preserving Denoising of Synthetic Aperture Radar Images
Synthetic aperture radar (SAR) is a widely used technology for acquiring landscape images from high altitude reconnaissance aircrafts or low-altitude spacecrafts. The acquired images, however, suffer from the effect of noise due to random phase fluctuations introduced in the signal during its acquisition. In this work we propose a noise removal algorithm which also maintains the texture of the acquired image.

Participating Organisations - IIT Patna India, INRIA Bordeaux France.
Area of Research - Geoscience and Remote Sensing.
Full Text - IEEE Geoscience and Remote Sensing Letters | IEEE Xplore  |  HAL INRIA  |  Bibtex

Joint Denoising-Deconvolution Approach for Fluorescence Microscopy
Image denoising and deconvolution are well known techniques applied to wide-field and confocal microscopy in order to restore images. The methods however suffer from their own drawbacks with denoising potentially resulting in smoothed images while deconvolution giving unpleasant artifacts due to ill-posedness of the problem. In this work, we propose to evaluate the interest of restoring the unknown image through a joint denoising and deconvolution program applied simultaneously in an iterative framework.

Participating Organisations - IIT Patna India, INSERM Paris, Institute Curie, Paris.
Area of Research - Medical Imaging.
Full Text - IEEE International Symposium on Biomedical Imaging ISBI 2016 | IEEE Xplore  |  Bibtex

Imaging Work-Flow for mRNA tracking in Living Yeast Cells
Objective is to quantify the newly transcribed mRNA behavior in living yeast cells and correlate transcriptional dynamics with nuclear pore complex (NPC) function. An adapted spinach-tagged transcript (green) is used to follow the dynamics of mRNA while the nuclear periphery is defined by tagging the nucleoporin Nup159 in mCherry (red). An imaging work-flow is then designed to address the biological questions (mRNA tracking, gene gating) simultaneously dealing with technical challenges of detecting little molecular entities and low signal-to-noise (SNR) images.

Participating Organisations - Vanderbuilt School of Medicine USA, CNRS France, INSERM Paris, Institute Curie, Paris.
Area of Research - Bioinformatics, Medical Imaging.
Full Text - Nature Communications | Nature.com  |  PubMed  |  Bibtex

Wavefront Phase Estimation Technique for Ground-Based Astronomical Telescopes
Turbulence in the Earth's atmosphere interferes with the propagation of planar wavefronts from outer space, resulting in a phase-distorted nonplanar wavefront. This phase distortion is responsible for the refractive blurring of images accounting to the loss in spatial resolution power of ground-based telescopes. Phase reconstruction from WFS measurements is done by solving large linear systems, followed by interpolating the low-resolution phase to its desired high resolution.

Participating Organisations - IIT Patna India, DOTA/E ONERA France, INRIA Bordeaux France.
Area of Research - Geoscience and Remote Sensing.
Full Text - IEEE Transactions in Geoscience and Remote Sensing | IEEE Xplore  |  HAL INRIA  |  Bibtex