Super-resolution Generative Adversarial Network Research Articles

Deep generative model super-resolves spatially correlated multiregional climate data

Super-resolving the coarse outputs of global climate simulations, termed downscaling, is crucial in making political and social decisions on systems requiring long-term climate change projections. Existing fast super-resolution techniques, however, have yet to preserve the spatially correlated nature of climatological data, which is particularly important when we address systems with spatial expanse, such as the development of transportation infrastructure. Herein, we show an adversarial network-based machine learning enables us to correctly reconstruct the inter-regional spatial correlations in downscaling with high magnification of up to 50 while maintaining pixel-wise statistical consistency. Direct comparison with the measured meteorological data of temperature and precipitation distributions reveals that integrating climatologically important physical information improves the downscaling performance, which prompts us to call this approach piSRGAN (Physics Informed Super-Resolution Generative Adversarial Network). The proposed method has a potential application to the inter-regionally consistent assessment of the climate change impact. Additionally, we present the outcomes of another variant of the deep generative model-based downscaling approach in which the low-resolution precipitation field is substituted with the pressure field, referred to as psiSRGAN (Precipitation Source Inaccessible SRGAN). Remarkably, this method demonstrates unexpectedly good downscaling performance for the precipitation field.

Applying machine learning to improve a texture type image

The paper is devoted to machine learning methods that focus on texture-type image enhancements, namely the improvement of objects in images. The aim of the study is to develop algorithms for improving images and to determine the accuracy of the considered models for improving a given type of images. Although currently used digital imaging systems usually provide high-quality images, external factors or even system limitations can cause images in many areas of science to be of low quality and resolution. Therefore, threshold values for image processing in a certain field of science are considered. The first step in image processing is image enhancement. The issues of signal image processing remain in the focus of attention of various specialists. Currently, along with the development of information technology, the automatic improvement of images used in any field of science is one of the urgent problems. Images were analyzed as objects: state license plates of cars, faces, sections of the field on satellite images. In this work, we propose to use the models of Super-Resolution Generative Adversarial Network (SRGAN), Extended Super-Resolution Generative Adversarial Networks (ERSGAN). For this, an experiment was conducted, the purpose of which was to retrain the trained ESRGAN model with three different architectures of the convolutional neural network, i.e. VGG19, MobileNet2V, ResNet152V2 to add perceptual loss (by pixels), also add more sharpness to the prediction of the test image, and compare the performance of each retrained model. As a result of the study, the use of convolutional neural networks to improve the image showed high accuracy, that is, on average ESRGAN+MobileNETV2 – 91 %, ESRGAN+VGG19 – 86 %, ESRGAN+ResNet152V2 – 96 %.

Application of generative adversarial neural networks for the formation of databases in scanning tunneling microscopy

We discuss the development of a technique for automatic generation of databases of images obtained with a scanning tunneling microscope. An analysis of state-of-the-art methods and means of automatic processing of images obtained from probe and electron microscopes is carried out. We proposed using generative-adversarial networks for generating images taken with a scanning tunneling microscope to form training databases of images. A process of training and comparison of deep convolutional generative adversarial network (DCGAN) architectures using the OpenCV and Keras libraries together with TensorFlow is described, with the best of them identified by computing the metrics IS, FID, and KID. The scaling of images obtained from DCGAN is performed using a method of fine tuning of a super-resolution generative adversarial neural network (SRGAN) and bilinear interpolation based on the Python programming language. An analysis of calculated quantitative metrics values shows that the best results of image generation are obtained using DCGAN96 and SRGAN. It is found that FID and KID metric values for SRGAN method are better than values for bilinear interpolation in all cases except for DCGAN32. All calculations are performed on a GTX GeForce 1070 video card. A method for automatic generation of a scanning tunneling microscope image database based on the stepwise application of DCGAN and SRGAN is developed. Results of generation and comparison of the original image, the one obtained with DCGAN96 and the enlarged image with SRGAN are presented.

ESRGAN-DP: Enhanced super-resolution generative adversarial network with adaptive dual perceptual loss

The proposal of perceptual loss solves the over-smoothing problem of images caused by pixel-wise loss and improves the visual quality of the images, but it also inevitably produces a large number of artifacts and distortions in images. The reason for this phenomenon is that the perceptual features only rely on a single pre-trained visual geometry group (VGG) network, which results in the features of the image being unable to be fully extracted, thus limiting the reasoning ability of the model. To fundamentally reduce the generation of artifacts and distortions, this paper proposes the Dual Perceptual Loss (DP Loss). First, we improve the perceptual feature extraction method so that it no longer only extracts single-type VGG features. In addition, a residual network (ResNet) feature that has a complementary relationship with the VGG feature can also be extracted. Then, we propose a dynamic weighting method to eliminate the magnitude difference between perceptual losses. Finally, to obtain the excellent effect of image reconstruction, enhanced super-resolution generative adversarial network (ESRGAN) with strong learning capability is used in this paper to adapt the complexity of DP Loss. The abundant experimental studies and evaluations are conducted on benchmark datasets. Results are encouraging and better than those previously reported on these datasets. The code is available at https://github.com/Sunny6-6-6/ESRGAN-DP.

Deep Learning-Based Prediction of Diabetic Retinopathy Using CLAHE and ESRGAN for Enhancement.

Vision loss can be avoided if diabetic retinopathy (DR) is diagnosed and treated promptly. The main five DR stages are none, moderate, mild, proliferate, and severe. In this study, a deep learning (DL) model is presented that diagnoses all five stages of DR with more accuracy than previous methods. The suggested method presents two scenarios: case 1 with image enhancement using a contrast limited adaptive histogram equalization (CLAHE) filtering algorithm in conjunction with an enhanced super-resolution generative adversarial network (ESRGAN), and case 2 without image enhancement. Augmentation techniques were then performed to generate a balanced dataset utilizing the same parameters for both cases. Using Inception-V3 applied to the Asia Pacific Tele-Ophthalmology Society (APTOS) datasets, the developed model achieved an accuracy of 98.7% for case 1 and 80.87% for case 2, which is greater than existing methods for detecting the five stages of DR. It was demonstrated that using CLAHE and ESRGAN improves a model's performance and learning ability.

Super-Resolution of Dental Panoramic Radiographs Using Deep Learning: A Pilot Study.

Using super-resolution (SR) algorithms, an image with a low resolution can be converted into a high-quality image. Our objective was to compare deep learning-based SR models to a conventional approach for improving the resolution of dental panoramic radiographs. A total of 888 dental panoramic radiographs were obtained. Our study involved five state-of-the-art deep learning-based SR approaches, including SR convolutional neural networks (SRCNN), SR generative adversarial network (SRGAN), U-Net, Swin for image restoration (SwinIr), and local texture estimator (LTE). Their results were compared with one another and with conventional bicubic interpolation. The performance of each model was evaluated using the metrics of mean squared error (MSE), peak signal-to-noise ratio (PNSR), structural similarity index (SSIM), and mean opinion score by four experts (MOS). Among all the models evaluated, the LTE model presented the highest performance, with MSE, SSIM, PSNR, and MOS results of 7.42 ± 0.44, 39.74 ± 0.17, 0.919 ± 0.003, and 3.59 ± 0.54, respectively. Additionally, compared with low-resolution images, the output of all the used approaches showed significant improvements in MOS evaluation. A significant enhancement in the quality of panoramic radiographs can be achieved by SR. The LTE model outperformed the other models.

On the modern deep learning approaches for precipitation downscaling

Deep Learning (DL) based downscaling has become a popular tool in earth sciences recently. Increasingly, different DL approaches are being adopted to downscale coarser precipitation data and generate more accurate and reliable estimates at local (~few km or even smaller) scales. Despite several studies adopting dynamical or statistical downscaling of precipitation, the accuracy is limited by the availability of ground truth. A key challenge to gauge the accuracy of such methods is to compare the downscaled data to point-scale observations which are often unavailable at such small scales. In this work, we carry out the DL-based downscaling to estimate the local precipitation data from the India Meteorological Department (IMD), which was created by approximating the value from station location to a grid point. To test the efficacy of different DL approaches, we apply four different methods of downscaling and evaluate their performance. The considered approaches are (i) Deep Statistical Downscaling (DeepSD), augmented Convolutional Long Short Term Memory (ConvLSTM), fully convolutional network (U-NET), and Super-Resolution Generative Adversarial Network (SR-GAN). A custom VGG network, used in the SR-GAN, is developed in this work using precipitation data. The results indicate that SR-GAN is the best method for precipitation data downscaling. The downscaled data is validated with precipitation values at IMD station. This DL method offers a promising alternative to statistical downscaling.

An improved generative adversarial network for remote sensing image super‐resolution

AbstractSpatial resolution is an important indicator that measures the quality of remote sensing images. Image texture has been successfully recovered by generative adversarial networks in deep learning super‐resolution (SR) methods. However, the existing methods are prone to image texture distortion. To solve the above problems, this paper proposes an improved generative adversarial network to enhance the super‐resolution reconstruction effect of medium‐ and low‐resolution (LR) remote sensing images. This network is based on the Super Resolution Generative Adversarial Network (SRGAN), which makes great improvements in the structure of the connection between the inside and outside of the residual block and the design of the model loss function. At the same time, the G1–G2–G3 structure between residuals effectively combines the image information of the three scales of small, medium and large. The model loss function can be designed based on the Charbonnier loss function to narrow the pixel distance between the reconstructed remote sensing image and the original image. Furthermore, targeted perceptual loss can direct the network to restore the texture details of the image according to the semantic category. The subjective and objective evaluation of the generated images and the ablation experiments prove that compared with SRGAN and other networks, our method can generate more realistic and reliable textures. Additionally, the indicators [peak signal‐to‐noise ratio (PSNR), structural similarity (SSIM), multiscale structural similarity (MS‐SSIM)] used to measure the quality of the reconstructed image obtain improved objective quantitative evaluation.

Large-Pore Network Simulations Coupled with Innovative Wettability Anchoring Experiment to Predict Relative Permeability of a Mixed-Wet Rock

Since the pioneering work of Oren et al. (SPE J 3(04):324–336, 1998), several attempts have been made to predict relative permeability curves with digital rock physics (DRP) technique. However, the problem has proved more complex than what researchers have expected, and these attempts failed. One of the main issues was the high number of uncertain parameters, especially for the wettability input, and this gets worst in mixed-wet scenario as the number of parameters is higher than in water-wet and oil-wet cases. In fact, Sorbie and Skauge (Petrophysics 53(06):401–409, 2012) stated that wettability assignment is the most complex and least validated stage in the DRP simulation workflow. Similarly, Bondino et al. (54(6):538–546, 2013) concluded that “genuine prediction” of multiphase flow properties will remain not credible until important progress is achieved in the area of wettability characterization at the pore scale. In this work, we propose a pragmatic approach to tackle these problems. First, we parallelize our pore network simulator in order to achieve large-scale PNM simulations. Then, we develop an innovative and fast anchoring experiment imaged by micro-CT scanner that helps to determine several wettability parameters needed for the DRP simulation (including the fraction of oil-wet/water-wet pores, any spatial or radius correlation of oil-wet pores, etc.). This experiment also provides an estimation of macroscopic parameters that help to anchor our pore-scale simulations and further reduce the uncertainty. In addition to help reducing the uncertainty of the simulation, this experiment provides a fast estimation of the wettability of the system. Images representing large volumes with low resolution are, first, improved with Enhanced Super-Resolution Generative Adversarial Networks (ESRGAN) to obtain a large image with high resolution. Then, a pore network is extracted, and TotalEnergies’ parallel pore network simulator is used for multiphase flow simulations considering the constraints from the anchoring experiment to reduce the uncertainty. Finally, we compare our simulations against high-quality SCAL experiment performed in-house and we assess the predictive power of our DRP workflow.

Deep learning based whale detection from satellite imagery

In spite of their fascinating and vulnerable status, the number of whales in the world's oceans remain debatable because of their living environment. As per statics, with the efforts made towards their conservation, there have been signs of recovery in the population of whales but still, many remote areas remain unsurveyed and thus the number of whales remain ambiguous. To maintain the balance of ecology it is needed to preserve the whales to balance the ecosystem and a key component of that will be to have a count of whales.To date, whale detection has been carried out using ship/aerial surveys, acoustic surveys, etc. but with the development in the field of Very High Resolution (VHR) satellite imagery, satellite images along with deep learning is also being effectively used for the detection of whales.Most of the previous work carried out for the detection of whales via satellite imagery is based upon the manual detection of whales by observers or scientists and that suffers from a high risk of false and inappropriate detections. There is also limited availability of satellite data in this field and there is a lack of usage of Artificial Intelligence for the automatic detection of whales.In this work we propose, a novel deep learning-based automatic whale detection approach from VHR satellite feed. We created a dataset of satellite images having evident chances of whales or probable whales by combining images from two different sources, used Enhanced Super-Resolution Generative Adversarial Networks (ESRGAN) to improve the quality of the images and maintain the natural texture of the images, augmented the captured satellite images to train the model, labelled the dataset as whale and probable whale. Finally, a whale dataset of 6000 whale satellite images is created which is the largest dataset available in this field. A deep learning-based whale detection approach is proposed to detect the existence of whales in the satellite image. The proposed approach performed better in terms of precision, recall, and mAP (Mean Average Precision). The precision achieved by the proposed system is 0.95, recall is 0.85, and mAP is 0.74 which is better as compared to the state-of-the-art work in this field.

Towards Large-Scale DRP Simulations: Generation of Large Super-Resolution images and Extraction of Large Pore Network Models

Representativity and accuracy of digital rock physics (DRP) simulations depend strongly on the size of the image volume and the resolution obtained. Even with one of the fastest DRP simulation techniques like pore network modelling, simulation volumes have typically been limited to few cubic millimetres for highly resolved images. In this paper, a super-resolution technique named enhanced super-resolution generative adversarial network (ESRGAN) is used to obtain well-resolved images with large field of view and to generate micro-CT images with resolution enhancement factors of × 4 and × 8. Subsets of resulting ESRGAN images were tested against the same volume (of acquisitions resolved at high and low resolution) by comparing petrophysical properties of interest. Pore network extraction and multiphase simulation results showed that pore size distribution, porosity, permeability, drainage capillary pressure and relative permeability curves obtained using ESRGAN images were more accurate. Large images, however, pose subsequent limitations on DRP simulations as pore network extraction code needs a lot of memory to process them (usually more than 60 GB of RAM for 15003 voxels image). Thus, we present a novel stitching strategy that is developed to enable the extraction of pore networks on such large images. Several validation cases of this method are presented to test the accuracy of the results from stitched networks on single- and multiphase flow properties. Finally, our stitching tool was used to generate two large networks of 3.6 million and 9.2 million elements, respectively, from two large ESRGAN images of approximately 49003 voxels.

A transformer-based synthetic-inflow generator for spatially developing turbulent boundary layers

This study proposes a newly developed deep-learning-based method to generate turbulent inflow conditions for spatially developing turbulent boundary layer (TBL) simulations. A combination of a transformer and a multiscale-enhanced super-resolution generative adversarial network is utilised to predict velocity fields of a spatially developing TBL at various planes normal to the streamwise direction. Datasets of direct numerical simulation (DNS) of flat plate flow spanning a momentum thickness-based Reynolds number, $Re_\theta = 661.5\unicode{x2013}1502.0$ , are used to train and test the model. The model shows a remarkable ability to predict the instantaneous velocity fields with detailed fluctuations and reproduce the turbulence statistics as well as spatial and temporal spectra with commendable accuracy as compared with the DNS results. The proposed model also exhibits a reasonable accuracy for predicting velocity fields at Reynolds numbers that are not used in the training process. With the aid of transfer learning, the computational cost of the proposed model is considered to be effectively low. Furthermore, applying the generated turbulent inflow conditions to an inflow–outflow simulation reveals a negligible development distance for the TBL to reach the target statistics. The results demonstrate for the first time that transformer-based models can be efficient in predicting the dynamics of turbulent flows. They also show that combining these models with generative adversarial networks-based models can be useful in tackling various turbulence-related problems, including the development of efficient synthetic-turbulent inflow generators.

Intelligent Image Super-Resolution for Vehicle License Plate in Surveillance Applications

Vehicle license plate images are often low resolution and blurry because of the large distance and relative motion between the vision sensor and vehicle, making license plate identification arduous. The extensive use of expensive, high-quality vision sensors is uneconomical in most cases; thus, images are initially captured and then translated from low resolution to high resolution. For this purpose, several traditional techniques such as bilinear, bicubic, super-resolution convolutional neural network, and super-resolution generative adversarial network (SRGAN) have been developed over time to upgrade low-quality images. However, most studies in this area pertain to the conversion of low-resolution images to super-resolution images, and little attention has been paid to motion de-blurring. This work extends SRGAN by adding an intelligent motion-deblurring method (termed SRGAN-LP), which helps to enhance the image resolution and remove motion blur from the given images. A comprehensive and new domain-specific dataset was developed to achieve improved results. Moreover, maintaining higher quantitative and qualitative results in comparison to the ground truth images, this study upscales the provided low-resolution image four times and removes the motion blur to a reasonable extent, making it suitable for surveillance applications.

The impact of load-PV profile resolution on distribution system risk assessment

Risk assessment is utilized to evaluate the risk level of the system and guide further operation and planning of utilities. However, to lower the cost of communication and storing, only a coarse resolution profile of load and PV can be collected and used to evaluate the risk level of the system, which may lead to inaccurate assessment. To this end, this paper investigates the impact of load-PV profile resolution on distribution system risk assessment. A super-resolution Generative Adversarial Network with pre- and post-processing processes is proposed to restore the high-frequency component of the load-PV profile. Then, with the generated high-resolution profile, a risk assessment model based on the exterior point method is proposed to evaluate the risk level of the distribution system. Case studies in a modified 33-bus system showed that the proposed super-resolution method could achieve the best performance to restore the peak values and high-frequency component of the load-PV profile compared to the other four methods. In addition, compared to super-resolution Generative Adversarial Networks without the post-processing process, the accuracy of risk assessment can be significantly improved using the proposed super-resolution method, which proves that the post-processing process is necessary.

Improving anisotropy resolution of computed tomography and annotation using 3D super-resolution network

In clinical practice, abdomen computed tomography (CT) is obtained with a slice thickness of 5 mm which causes the anisotropy resolution. Especially in 3D automatic medical segmentation tasks, this anisotropy resolution in transverse plane and z-axial of CT image causes the unbalance of spatial feature for 3D convolution, which further limits the quality of segmentation. To recover context features between slices, super-resolution networks can better reconstruct detail information than interpolation methods. To reconstruct CT from different scanner models, a stronger generalization ability is indispensable. Moreover, to improve segmentation performance, the annotation of lesion area should be reconstructed at the same time. To address these issues, an Average Super-Resolution Generative Adversarial Network (ASRGAN) is proposed in this paper. We designed a multi-path average block to recover inter-slice information from CT with different image quality. Experimental results demonstrate that the proposed ASRGAN is superior to other methods on reconstruction with 2.42 db improvement on PSNR. And based on its reconstruction results, it further promotes improving 3D segmentation of the abdominal lesion liver tumor by 4.00% and the abdominal viscera pancreas by 2.25% on dice, to further reveal the effects of our reconstruction from view of this follow-up medical image analysis.

Skin Cancer Classification Framework Using Enhanced Super Resolution Generative Adversarial Network and Custom Convolutional Neural Network

Melanin skin lesions are most commonly spotted as small patches on the skin. It is nothing but overgrowth caused by melanocyte cells. Skin melanoma is caused due to the abnormal surge of melanocytes. The number of patients suffering from skin cancer is observably rising globally. Timely and precise identification of skin cancer is crucial for lowering mortality rates. An expert dermatologist is required to handle the cases of skin cancer using dermoscopy images. Improper diagnosis can cause fatality to the patient if it is not detected accurately. Some of the classes come under the category of benign while the rest are malignant, causing severe issues if not diagnosed at an early stage. To overcome these issues, Computer-Aided Design (CAD) systems are proposed which help to reduce the burden on the dermatologist by giving them accurate and precise diagnosis of skin images. There are several deep learning techniques that are implemented for cancer classification. In this experimental study, we have implemented a custom Convolution Neural Network (CNN) on a Human-against-Machine (HAM10000) database which is publicly accessible through the Kaggle website. The designed CNN model classifies the seven different classes present in HAM10000 database. The proposed experimental model achieves an accuracy metric of 98.77%, 98.36%, and 98.89% for protocol-I, protocol-II, and protocol-III, respectively, for skin cancer classification. Results of our proposed models are also assimilated with several different models in the literature and were found to be superior than most of them. To enhance the performance metrics, the database is initially pre-processed using an Enhanced Super Resolution Generative Adversarial Network (ESRGAN) which gives a better image resolution for images of smaller size.

4G/VoLTE- Spectrum Sensing using Cyclostationary Detection (CS-SS) Method: A Review

Artificial Neural Networks (ANNs), Deep Learning, and AI Deep learning is a subset of machine learning. In order to create pictures with a greater resolution, a high-resolution GAN combines a deep network with an opponent network. An approach to generative modelling that uses deep learning techniques like convolutional neural networks is known as generative adversarial networks, or GAN. In the Super-Resolution Generative Adversarial Network (SRGAN), a Generative Adversarial Network (GAN) may transform low-resolution images into super resolution images that are more finely detailed and of higher quality. In the past, CNNs were used to produce incredibly precise and detailed images. However, they could have trouble recalling specifics and usually draw hazy visuals which can be overcome by SRGAN.

A Review Super Resolution Using Generative Adversarial Network-Applications and Challenges

Artificial Neural Networks (ANNs), Deep Learning, and AI Deep learning is a subset of machine learning. In order to create pictures with a greater resolution, a high-resolution GAN combines a deep network with an opponent network. An approach to generative modelling that uses deep learning techniques like convolutional neural networks is known as generative adversarial networks, or GAN. In the Super-Resolution Generative Adversarial Network (SRGAN), a Generative Adversarial Network (GAN) may transform low-resolution images into super resolution images that are more finely detailed and of higher quality. In the past, CNNs were used to produce incredibly precise and detailed images. However, they could have trouble recalling specifics and usually draw hazy visuals which can be overcome by SRGAN.

Prostatic urinary tract visualization with super-resolution deep learning models.

In urethra-sparing radiation therapy, prostatic urinary tract visualization is important in decreasing the urinary side effect. A methodology has been developed to visualize the prostatic urinary tract using post-urination magnetic resonance imaging (PU-MRI) without a urethral catheter. This study investigated whether the combination of PU-MRI and super-resolution (SR) deep learning models improves the visibility of the prostatic urinary tract. We enrolled 30 patients who had previously undergone real-time-image-gated spot scanning proton therapy by insertion of fiducial markers. PU-MRI was performed using a non-contrast high-resolution two-dimensional T2-weighted turbo spin-echo imaging sequence. Four different SR deep learning models were used: the enhanced deep SR network (EDSR), widely activated SR network (WDSR), SR generative adversarial network (SRGAN), and residual dense network (RDN). The complex wavelet structural similarity index measure (CW-SSIM) was used to quantitatively assess the performance of the proposed SR images compared to PU-MRI. Two radiation oncologists used a 1-to-5 scale to subjectively evaluate the visibility of the prostatic urinary tract. Cohen's weighted kappa (k) was used as a measure of agreement of inter-operator reliability. The mean CW-SSIM in EDSR, WDSR, SRGAN, and RDN was 99.86%, 99.89%, 99.30%, and 99.67%, respectively. The mean prostatic urinary tract visibility scores of the radiation oncologists were 3.70 and 3.53 for PU-MRI (k = 0.93), 3.67 and 2.70 for EDSR (k = 0.89), 3.70 and 2.73 for WDSR (k = 0.88), 3.67 and 2.73 for SRGAN (k = 0.88), and 4.37 and 3.73 for RDN (k = 0.93), respectively. The results suggest that SR images using RDN are similar to the original images, and the SR deep learning models subjectively improve the visibility of the prostatic urinary tract.

(Retracted) Super-resolution generative adversarial networks using autoencoder reduce dimension

Generative adversarial network (GAN) has been widely applied to produce super-resolution (SR) image with real perception and texture details. In most existing SR approaches, the training objective typically measures a pixel-wise average distance between the SR and high-resolution (HR) images. However, as the degradation function of different images from HR to low resolution (LR) is generally different, optimizing such metrics often leads to certain unpleasant artificial traces. Unlike the prevalent GAN inversion methods that require expensive image-specific optimization at runtime, we present an alternative formulation by directly leveraging latent representation produced by a pretrained AutoEncoder. We call this improved method reduce dimension super-resolved GAN (RD-SRGAN). RD-SRGAN first obtains the latent feature representation of LR image by a pretrained AutoEncoder as input to the generator network. This process not only reduces noise effects but also decreases the overall computational complexity. On the other hand, the residual between the ground truth and the produced images replaces the produced images as input to the discriminator network, and a 2D zero mean Gaussian noise with controllable low variance replaces the real images as another input to the discriminator network. By leveraging the feature representation and properties of the 2D zero mean Gaussian noise, we restrict the optimization space to produce an SR image. Therefore, the residual of the generated SR images tend to approximates to a Gaussian noise, which introduces useless deviation information as little as possible. Experimental results show that RD-SRGAN can benefit from these strategies and achieve improved fidelity and naturalness comparison to existing methods. Switching the pretrained AutoEncoder allows the method to deal with images from diverse categories, e.g., remote sensing satellite imaging, medical imaging, and astronomy.