Search

The development of telescopes with large apertures is driven by the increasing demand for higher imaging resolution and coverage. Large membrane diffraction space telescopes are particularly attractive due to their lightweight nature, ease of folding and unfolding, and high-precision imaging capabilities. However, their substantial size and flexibility make them susceptible to long-duration, low-frequency vibrations during attitude maneuvers, which degrade attitude and imaging accuracy and risk instrument damage. Consequently, an urgent need exists for a high-precision integrated control scheme. In this paper, the integrated control of vibration and attitude in a large space telescope is studied using cable actuators and control moment gyroscopes (CMGs). Since cables can only withstand limited tension, the unilateral and constrained characteristics of the control input are taken into account during the control design process. Firstly, a rigid-flexible coupling dynamic model of the space telescope is established using the velocity variation principle with hybrid coordinates. Additionally, a two-body astrodynamic model is developed to assess the impact of gravity gradient torque. Next, combining the computational torque method and H∞ control theory, an integrated control scheme is proposed, which achieves vibration and attitude control during maneuvers. Then, this paper optimizes the cable actuator positions via the discrete particle swarm optimization (PSO) algorithm and plans various attitude maneuver paths. Finally, numerical simulations are adopted to demonstrate the effectiveness of the control scheme. The results show that this integrated control scheme swiftly suppresses structural vibrations during attitude maneuvers, enabling high-precision attitude control of the space telescope. • A vibration-attitude integrated control for a novel large membrane diffraction space telescope is studied • 3 types of attitude maneuver paths are planned for the two requirements of time optimal and minimum vibration excitation. [ABSTRACT FROM AUTHOR]

Training deep neural networks for semantic segmentation of aerial images relies heavily on obtaining a large number of precise pixel-level annotations, which can cause significant annotation expenses. Given the fact that acquiring fine-class annotations is considerably more challenging than obtaining coarse-class annotations, we present a novel semi-supervised learning framework, which utilizes high spatial resolution images annotated with coarse-class labels alongside a very small set of fine-grained annotated images as the training set, thereby achieving classification results that are refined in both spatial resolution and categorical granularity. Specifically, this framework adopts Mix Transformer (MiT) as the backbone architecture to accommodate both local feature extraction and long-range dependency modeling capabilities and utilizes multi-prototype learning to model each class as multiple sub-prototypes, preserving the intrinsic variance characteristics within classes. We propose a dedicated co-training approach tailored for extracting fine-grained pseudo-labels from coarse-grained samples. In this approach, a local-softmax pseudo-labeling strategy is developed to ensure a harmonious balance between the efficiency and accuracy of the pseudo-labeling, and four losses are formulated for both single-level class and cross-category granularity supervised learning. We evaluate the proposed framework on the Gaofen Image Dataset (GID) and Five-Billion-Pixels (FBP) dataset, confirming its feasibility and superior results. In particular, based on coarse-class annotations, the performance achieved using only 5% of fine-class labels, in terms of the four metrics, namely mIoU, mean UA, mean F1-score, and OA, reached 91%, 96%, 89%, and 93% of the fully-supervised baseline performance respectively. The code is available at https://github.com/chenhaocs/C2F. [ABSTRACT FROM AUTHOR]

To develop acceleration strategies for 3D multi-slab diffusion weighted imaging (3D ms-DWI) for enabling applications that require simultaneously high spatial (1 mm isotropic) and angular (> 30 directions) resolutions. 3D ms-DWI offers high SNR-efficiency, with the ability to achieve high isotropic spatial resolution, yet suffers from long scan-times for studies requiring high angular resolutions. We develop 6D k-q space acceleration strategies to reduce the scan-time. Specifically, we develop non-uniform 3D k y - k z under-sampling employing a shot-selective 2D CAIPI sampling approach. To achieve inter-shot phase-compensation, 2D navigators were employed that utilize the same CAIPI trajectory. An iterative model-based 3D multi-shot reconstruction was designed by incorporating phase into the forward encoding process. Additionally, the shot-selective non-uniform k y - k z CAIPI acceleration was randomized along the q-dimension. The 3D model-based multi-shot reconstruction is then extended to a joint reconstruction that simultaneously reconstructs all the q-space points, with the help of a spatial total variation and deep-learned q-space regularization. The proposed reconstruction is shown to achieve adequate phase-compensation in both 2D CAIPI accelerated and additional k y - k z under-sampled cases. Using retrospective under-sampling experiments, we show that k-q accelerations close a factor of 12 can be achieved with a reconstruction error < 3% for both single and multi-shell data. This translates to a scan-time reduction by 3-fold for experiments with simultaneously high spatial and angular resolutions. The proposed method facilitates the utilization of 3D ms-DWI for simultaneously high k-q resolution applications with close to 3× reduced scan-time. [ABSTRACT FROM AUTHOR]

Bone diseases deteriorate the microstructure of bone tissue. Optical-resolution photoacoustic microscopy (OR-PAM) enables high spatial resolution of imaging bone tissues. However, the spatiotemporal trade-off limits the application of OR-PAM. The purpose of this study was to improve the quality of OR-PAM images without sacrificing temporal resolution. In this study, we proposed the Photoacoustic Dense Attention U-Net (PADA U-Net) model, which was used for reconstructing full-scanning images from under-sampled images. Thereby, this approach breaks the trade-off between imaging speed and spatial resolution. The proposed method was validated on resolution test targets and bovine cancellous bone samples to demonstrate the capability of PADA U-Net in recovering full-scanning images from under-sampled OR-PAM images. With a down-sampling ratio of [4, 1], compared to bilinear interpolation, the Peak Signal-to-Noise Ratio and Structural Similarity Index Measure values (averaged over the test set of bovine cancellous bone) of the PADA U-Net were improved by 2.325 dB and 0.117, respectively. The results demonstrate that the PADA U-Net model reconstructed the OR-PAM images well with different levels of sparsity. Our proposed method can further facilitate early diagnosis and treatment of bone diseases using OR-PAM. [ABSTRACT FROM AUTHOR]

Luminescence imaging in the second near-infrared (NIR-II) window, particularly within the NIR-IIb (1500–1700 nm) sub-window, has been utilized for both fundamental research and clinical applications due to its exceptional spatiotemporal resolution. Nevertheless, efficient contrast agents designed for NIR-IIb imaging are still scarce. Herein, a multilayered rare-earth nanocrystal (NaYF 4 :Yb,Er,Ce@NaYF 4 :Yb,Nd@NaYF 4) with strong NIR-IIb luminescence was designed and synthesized. The doping of Ce3+ ions in the core effectively suppresses the upconversion and boosts the downconversion luminescence by populating the Er 4I 13/2 level. Additionally, by separating the sensitizer (Nd3+) and the activator (Er3+) in different layers, we are able to maximize the light absorption around 800 nm and reduce the cross relaxations between rare-earth elements, which leads to a strong downconversion emission around 1530 nm under 808 nm excitation. After passivation with NaYF 4 shell and phase transfer with highly PEGylated mesoporous silica layer, the nanocomposites with excellent colloidal stability and bright NIR-IIb emission were obtained. With long blood circulation time in vivo, the fabricated nanocomposites enable high resolution and contrast luminescence imaging of the blood vessels and tumors in living mice. Moreover, the rare-earth nanocrystals with superior optical properties also hold great promise for various applications in optical sensing and biomedical imaging. [Display omitted] [ABSTRACT FROM AUTHOR]

In this paper, cubic-phase Lu 2 O 3 :Bi3+ nanoparticles, with a range of sizes between 110 and 180 nm, were obtained by a combination of co-precipitation and thermal annealing techniques. For these nanoparticles upon UV or X-ray excitation, strong broadband emission peaks at 410 and 480 nm corresponding to Bi3+ occupying S 6 and C 2 (Lu3+(S 6) and Lu3+(C 2)) sites, respectively, can be observed. On the basis of the concentration-dependent luminescence intensity ratios (I 480 (C 2)/I 410 (S 6)) of Bi3+(C 2) to Bi3+(S 6), the energy transfer from Bi3+(S 6) to Bi3+(C 2) can be confirmed at high Bi3+ concentration. In particular, their X-ray absorption coefficient, emission wavelength, afterglow level (60 ppm) and response time (465 ns) are almost comparable to those of the commercial Bi 4 Ge 3 O 12 crystals. By the means of dispersing Lu 1.98 Bi 0.02 O 3 nanoscintillators into organic polymer (PMMA)-acetone solution, the Lu 1.98 Bi 0.02 O 3 -PMMA flexible composite films exhibiting rapid response to X-ray photons, high X-ray sensitivity and radiation resistance, good mechanical performance and excellent environmental stability were prepared using a spin-coating technique. On the surface of X-ray imaging screen formed by the flexible dense composite films, the spatial resolution can reach to 6.0 lp/mm at a safe irradiation dose (4.6 μGy). Meanwhile, for the imaging screen using the flexible composite films after bended for thousands of times, we can observe clear image of each reed gap (300 μm) of the music box. For the Lu 2 O 3 :Bi3+-PMMA composite films, our results suggest that not only their scintillation properties are almost comparable to those of the commercial broadband Bi 4 Ge 3 O 12 crystals, but also they might be applied in large-area and non-planar X-ray imaging. [ABSTRACT FROM AUTHOR]

• The potential of advanced imaging techniques and processing image methods to recover unreadable text in historic documents affected by chemical damage, as a consequence of the application of ink enhancement reagents in the 19th century ("galling"), was demonstrated. • High resolution multiband imaging (MBI) techniques, namely UV-induced visible fluorescence (UVL) and UV reflected (UVR) imaging, were efficient for the recovery of unreadable text in documents with faded ink, soiled parchment and moderate localised galling. • The contents of documents affected by severe chemical damage were recoverable by means of reflectance imaging spectroscopy (RIS) in the VNIR and micro X ray fluorescence imaging. • Image processing methods such as PCA or elemental map addition and subtraction were key for an increased readability of the text in the chemically damaged documents. • Combining these techniques, the legibility of 31 documents from the collection of the national archives of the UK could be significantly improved, providing new relevant information for the medieval history of Ireland. From fires to floods, from invisible inks to redactions, information has been accidentally or intentionally obscured on countless documents, maps and photographs, in collections from archives, libraries, and museums across the world. Removed from catalogues and reading rooms and therefore inaccessible to researchers and the general public, what new knowledge would these obscured documents reveal if we were able to read them? This paper demonstrates that current imaging technologies can be used to unlock this lost content. We used High Resolution Multiband Imaging (MBI), Reflectance Imaging Spectroscopy (RIS) and micro X Ray Fluorescence (µXRF) imaging to reveal illegible letter-forms and whole words written in iron gall ink on parchment from medieval documents relevant to the 'Virtual Record Treasury of Ireland' project, a digital reconstruction of archives destroyed in 1922 at the Public Record Office of Ireland at the outset of the Irish Civil War. The readability of text in historic documents with faded iron gall ink and/or parchment obscured by damp stains and ingrained dirt was successfully enhanced with MBI techniques, such as UV-induced visible luminescence (UVL) and UV reflected (UVR) imaging, which were able to recover the contrast of the ink. However, visualising text in documents showing chemical damage caused by the application of reagents for ink enhancement in the 19th century (a practice known as "galling") was more challenging because of the similar composition of the ink and the staining. RIS proved helpful to read documents with moderate to severe chemical damage, particularly after applying Principal Component Analysis (PCA). Text in documents that were severely damaged by galling was also recoverable with µXRF imaging of the distribution of iron from the ink. This also posed several challenges, such as minimising movement of the parchment during overnight scans and separating the data corresponding to iron on the front of the parchment which was combined with the signal from the back. Through subtraction of elemental maps during post-processing, we managed to tackle the latter issue. The combination of these different imaging techniques enabled palaeographers and medieval records specialists to identify individual characters and whole words, thereby recovering the meaning of texts that were previously indecipherable. Further research will adapt the methodology to the broad range of causes of information loss, including other reagents historically used for ink enhancement, as well as the variety of document media and temporalities, which require an extensive and multi-faceted approach of advanced imaging and post-processing techniques, in balance with the preservation needs of these fragile and invaluable historic materials. [ABSTRACT FROM AUTHOR]

This study aimed to realise 3-D super-resolution ultrasound imaging transcutaneously with a row-column array which has far fewer independent electronic channels and a wider field of view than typical fully addressed 2-D matrix arrays. The in vivo image quality of the row-column array is generally poor, particularly when imaging non-invasively. This study aimed to develop a suite of image formation and post-processing methods to improve image quality and demonstrate the feasibility of ultrasound localisation microscopy using a row-column array, transcutaneously on a rabbit model and in a human. To achieve this, a processing pipeline was developed which included a new type of rolling window image reconstruction, which integrated a row-column array specific coherence-based beamforming technique with acoustic sub-aperture processing. This and other processing steps reduced the 'secondary' lobe artefacts, and noise and increased the effective frame rate, thereby enabling ultrasound localisation images to be produced. Using an in vitro cross tube, it was found that the procedure reduced the percentage of 'false' locations from ∼26% to ∼15% compared to orthogonal plane wave compounding. Additionally, it was found that the noise could be reduced by ∼7 dB and the effective frame rate was increased to over 4000 fps. In vivo, ultrasound localisation microscopy was used to produce images non-invasively of a rabbit kidney and a human thyroid. It has been demonstrated that the proposed methods using a row-column array can produce large field of view super-resolution microvascular images in vivo and in a human non-invasively. [ABSTRACT FROM AUTHOR]

Magnetic resonance imaging (MRI) is non-invasive and crucial for clinical diagnosis, but it has long acquisition time and aliasing artifacts. Accelerated imaging techniques can effectively reduce the scanning time of MRI, thereby decreasing the anxiety and discomfort of patients. Vision Transformer (ViT) based methods have greatly improved MRI image reconstruction, but their computational complexity and memory requirements for the self-attention mechanism grow quadratically with image resolution, which limits their use for high resolution images. In addition, the current generative adversarial networks in MRI reconstruction are difficult to train stably. To address these problems, we propose a Local Vision Transformer (LVT) based adversarial Diffusion model (Diff-GAN) for accelerating MRI reconstruction. We employ a generative adversarial network (GAN) as the reverse diffusion model to enable large diffusion steps. In the forward diffusion module, we use a diffusion process to generate Gaussian mixture distribution noise, which mitigates the gradient vanishing issue in GAN training. This network leverages the LVT module with the local self-attention, which can capture high-quality local features and detailed information. We evaluate our method on four datasets: IXI, MICCAI 2013, MRNet and FastMRI, and demonstrate that Diff-GAN can outperform several state-of-the-art GAN-based methods for MRI reconstruction. [ABSTRACT FROM AUTHOR]

To evaluate the performance of high-resolution free-breathing (FB) hepatobiliary phase imaging of the liver using the eXtra-Dimension Golden-angle RAdial Sparse Parallel (XD-GRASP) MRI technique. Fifty-eight clinical patients (41 males, mean age = 52.9 ± 12.9) with liver lesions who underwent dynamic contrast-enhanced MRI with a liver-specific contrast agent were prospectively recruited for this study. Both breath-hold volumetric interpolated examination (BH-VIBE) imaging and FB imaging were performed during the hepatobiliary phase. FB images were acquired using a stack-of-stars golden-angle radial sequence and were reconstructed using the XD-GRASP method. Two experienced radiologists blinded to acquisition schemes independently scored the overall image quality, liver edge sharpness, hepatic vessel clarity, conspicuity of lesion, and overall artifact level of each image. The non-parametric paired two-tailed Wilcoxon signed-rank test was used for statistical analysis. Compared to BH-VIBE images, XD-GRASP images received significantly higher scores (P < 0.05) for the liver edge sharpness (4.83 ± 0.45 vs 4.29 ± 0.46), the hepatic vessel clarity (4.64 ± 0.67 vs 4.15 ± 0.56) and the conspicuity of lesion (4.75 ± 0.53 vs 4.31 ± 0.50). There were no significant differences (P > 0.05) between BH-VIBE and XD-GRASP images for the overall image quality (4.61 ± 0.50 vs 4.74 ± 0.47) and the overall artifact level (4.13 ± 0.44 vs 4.05 ± 0.61). Compared to conventional BH-VIBE MRI, FB radial acquisition combined with XD-GRASP reconstruction facilitates higher spatial resolution imaging of the liver during the hepatobiliary phase. This enhancement can significantly improve the visualization and evaluation of the liver. [ABSTRACT FROM AUTHOR]

Increasing the signal-to-noise ratio (SNR) has always been of critical importance for magnetic resonance imaging. Although increasing field strength provides a linear increase in SNR, it is more and more costly as field strength increases. Therefore, there is a major effort today to use signal processing methods to improve SNR since it is more efficient and economical. There are a variety of methods to improve SNR such as averaging the data at the expense of imaging time, or collecting the data with a lower resolution, all of these methods, including imaging processing methods, usually come at the expense of loss of image detail or image blurring. Therefore, we developed a new mathematical approach called CROWN (C onstrained R econstruction o f W hite N oise) to enhance SNR without loss of structural detail and without affecting scanning time. In this study, we introduced and tested the concept behind CROWN specifically for STAGE (strategically acquired gradient echo) imaging. The concept itself is presented first, followed by simulations to demonstrate its theoretical effectiveness. Then the SNR improvement on proton spin density (PSD) and R2⁎ maps was investigated using brain STAGE data acquired from 10 healthy controls (HCs) and 10 patients with Parkinson's disease (PD). For the PSD and R2* maps, the SNR and CNR between white matter and gray matter were improved by a factor of 1.87 ± 0.50 and 1.72 ± 0.88, respectively. The white matter hyperintensity lesions in PD patients were more clearly defined after CROWN processing. Using these improved maps, simulated images for any repeat time, echo time or flip angle can be created with improved SNR. The potential applications of this technology are to trade off the increased SNR for higher resolution images and/or faster imaging. • Introduce a new mathematical approach called CROWN to enhance the image SNR without loss of structural detail. • The concept was explained and tested using simulations and brain STAGE data for healthy controls and Parkinson's patients. • In general, the SNR can be improved by a factor of 1.87 ± 0.50. [ABSTRACT FROM AUTHOR]

Small molecule drugs play a pivotal role in the arsenal of anticancer pharmacological agents. Nonetheless, their small size poses a challenge when directly visualizing their localization, distribution, mechanism of action (MOA), and target engagement at the subcellular level in real time. We propose a strategy for developing triple-functioning drug beacons that seamlessly integrate therapeutically relevant bioactivity, precise subcellular localization, and direct visualization capabilities within a single molecular entity. As a proof of concept, we have meticulously designed and constructed a boronic acid fluorescence drug beacon using coumarin–hemicyanine (CHB). Our CHB design includes three pivotal features: a boronic acid moiety that binds both adenosine triphosphate (ATP) and adenosine diphosphate (ADP), thus depleting their levels and disrupting the energy supply within mitochondria; a positively charged component that targets the drug beacon to mitochondria; and a sizeable conjugated luminophore that emits fluorescence, facilitating the application of structured illumination microscopy (SIM). Our study indicates the exceptional responsiveness of our proof-of-concept drug beacon to ADP and ATP, its efficacy in inhibiting tumor growth, and its ability to facilitate the tracking of ADP and ATP distribution around the mitochondrial cristae. Furthermore, our investigation reveals that the micro-dynamics of CHB induce mitochondrial dysfunction by causing damage to the mitochondrial cristae and mitochondrial DNA. Altogether, our findings highlight the potential of SIM in conjunction with visual drug design as a potent tool for monitoring the in situ MOA of small molecule anticancer compounds. This approach represents a crucial advancement in addressing a current challenge within the field of small molecule drug discovery and validation. Drug beacon CHB inhibits energy production by binding to ATP/ADP in mitochondria, disrupting the structure and function of mitochondria, exerting anti-tumor effects. [Display omitted] [ABSTRACT FROM AUTHOR]

High resolution (HR) images consist of higher quality and more detail information in comparison to low-resolution images. But obtaining HR images entails higher costs and requires a larger workforce. The solution is using super-resolution (SR) images. Single image SR is the process of creating HR images from a single low-resolution image. Obtaining SR images from low-resolution images is a well-known problem in the domains of computer vision and image processing. Advancement in technology has given rise to many SR algorithms. The perceptual quality assessment of super-resolved images can be used to benchmark the techniques employed for image SR. In this work, a cascaded network-blind super-resolution image quality assessment (CN-BSRIQA) methodology is proposed. The proposed approach works under the cascaded architecture where a convolutional neural network (CNN) is cascaded with a deep belief network (DBN). CNN is employed as a shallow network in the proposed methodology to extract low-level information after partitioning the input image into patches. To assess the visual quality of the SR images, the features retrieved from CNN are incorporated into a DBN. Three databases are used to evaluate the performance of proposed CN-BSRIQA i.e. , SR Quality Database (SRQD), SR Image Quality Database (SRID), and visual quality evaluation for super-resolved images (QADS). When compared to other state-of-the-art methodologies for assessing the visual quality of SR images, CN-BSRIQA outperforms them. For the perceptual quality assessment of SR images, the experimental results reveal that CNN-based techniques outperform techniques based on hand-crafted features. Furthermore, the shallow CNN proposed in the CN-BSRIQA can extract features that are content-independent i.e. , they show better performance over cross-database evaluation in comparison to existing state-of-the-art techniques. [ABSTRACT FROM AUTHOR]

• Fe/Mg phyllosilicates and mono- and polyhydrated Mg sulfates are detected in Mikumi crater. • Possible desiccation/hydraulic fracturing on crater floor. • Phyllosilicates are on top of the sulfate units, meaning that they formed later than sulfates. • Mikumi crater is a key area in Meridiani Planum to investigate Mars' ancient climate. The region of Meridiani Planum on Mars retains the evidence of multiple surface and near-surface aqueous processes that span from the Noachian to the Hesperian. This makes it an interesting spot for better understanding the characteristics and the evolution of Mars' aqueous environment between these two fundamental epochs. Here we investigate the mineralogy, stratigraphy, and morphology of the Noachian–Hesperian sedimentary materials that fill a 16-km-wide crater named Mikumi, located in the northern part of Meridiani Planum. The analysis was carried out combining observations from the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM), the High Resolution Imaging Science Experiment (HiRISE), and the Context Imager (CTX). About 20–30 meters, on average, of Fe/Mg phyllosilicate-rich rocks and a thick sequence (75 m) of both mono- and polyhydrated Mg sulfates are detected on the crater floor. Both layers postdate the impact event. Most of the hydrated terrains are extensively polygonally fractured, suggesting the possible action of water-related processes such as desiccation and hydro fracturing on the sediments. Stratigraphic analysis clearly shows that Fe/Mg phyllosilicates were deposited on top of the sulfate sequence, meaning that they are younger than sulfates. This can have interesting implications on paleoclimatic studies of Mars as it shows that a simple timeline going from a clay-rich Noachian to a sulfate-rich Hesperian oversimplifies the evolution of Mars' mineralogy and related aqueous environment at the Noachian–Hesperian boundary. [ABSTRACT FROM AUTHOR]

The development of super-resolution fluorescence microscopy is very essential for understanding the physical and biological fundamentals at nanometer scale. However, to date most super-resolution modalities require either complicated/costly purpose-built systems such as multiple-beam architectures or complex post-processing procedures with intrinsic artifacts. Achieving three-dimensional (3D) or multi-channel sub-diffraction microscopic imaging using a simple method remains a challenging and struggling task. Herein, we proposed 3D highly-nonlinear super-resolution microscopy using a single-beam excitation strategy, and the microscopy principle was modelled and studied based on the ultrahigh nonlinearity enabled by photon avalanches. According to the simulation, the point spread function of highly nonlinear microscopy is switchable among different modes and can shrink three-dimensionally to sub-diffraction scale at the photon avalanche mode. Experimentally, we demonstrated 3D optical nanoscopy assisted with huge optical nonlinearities in a simple laser scanning configuration, achieving a lateral resolution down to 58 nm (λ /14) and an axial resolution down to 185 nm (λ /5) with one single beam of low-power, continuous-wave, near-infrared laser. We further extended the photon avalanche effect to many other emitters to develop multi-color photon avalanching nanoprobes based on migrating photon avalanche mechanism, which enables us to implement single-beam dual-color sub-diffraction super-resolution microscopic imaging. [ABSTRACT FROM AUTHOR]

Deep convolutional neural networks (DCNNs) have exhibited excellent feature extraction and detail reconstruction capabilities for single image super-resolution (SISR). Nevertheless, most previous DCNN-based methods do not fully utilize the complementary strengths between feature maps, channels, and pixels. Therefore, it hinders the ability of DCNNs to represent abundant features. To tackle the aforementioned issues, we present a Cascaded Visual Attention Network for SISR called CVANet, which simulates the visual attention mechanism of the human eyes to focus on the reconstruction process of details. Specifically, we first designed a trainable feature attention module (FAM) for feature-level attention learning. Afterward, we introduce a channel attention module (CAM) to reinforce feature maps under channel-level attention learning. Meanwhile, we propose a pixel attention module (PAM) that adaptively selects representative features from the previous layers, which are utilized to generate a high-resolution image. Satisfactory, our CVANet can effectively improve the resolution of images by exploring the feature representation capabilities of different modules and the visual perception properties of the human eyes. Extensive experiments with different methods on four benchmarks demonstrate that our CVANet outperforms the state-of-the-art (SOTA) methods in subjective visual perception, PSNR, and SSIM.The code will be made available https://github.com/WilyZhao8/CVANet. [ABSTRACT FROM AUTHOR]

Recently, many super-resolution (SR) methods based on convolutional neural networks (CNNs) have achieved superior performance by utilizing deep and heavy models, which may not be suitable for real-world low-budget devices. To address this issue, we propose a novel lightweight SR network called a multi-scale feature selection network (MFSN). As the basic building block of MFSN, the multi-scale feature selection block (MFSB) is presented to extract the rich multi-scale features from a coarse-to-fine receptive field level. For a better representation ability, a wide-activated residual unit is adopted in each branch of MFSB except the last one. In MFSB, the scale selection module (SSM) is designed to effectively fuse the features from two adjacent branches by adjusting receptive field sizes adaptively. Further, a comprehensive channel attention mechanism (CCAM) is integrated into SSM to learn the dynamic selection weight by considering the local and global inter-channel dependencies. Extensive experimental results illustrate that the proposed MFSN is superior to other lightweight methods. • The MFSB module is proposed to extract multi-scale features from a coarse-to-fine receptive field level. • The SSM module is designed to well fuse multi-scale features by adjusting the receptive field adaptively. • The CCAM mechanism is built in the SSM to learn the selection weights by considering the global and local inter-channel dependencies. • The proposed MFSN is more efficient and effective than many lightweight CNN-based SR methods. [ABSTRACT FROM AUTHOR]

The image super-resolution (SR) operation holds multiple solutions with the one-to-many mapping from low-resolution (LR) to high-resolution (HR) space. However, the SR of different scales for the same image is usually regarded as independent tasks in the existing SR networks. Therefore, these networks are inflexible to effectively utilize feature learning experience and require much more computing time to recover HR images in higher resolutions. Recent arbitrary scale SR methods still cannot solve these problems. To efficiently and effectively recover HR images, this paper presents an efficient multi-scale learning method for image SR networks based on a novel self-generating (SG) mechanism. This method (briefly named SG-SR) utilizes the feature learning results of SR networks to generate upscale filters by using the novel SG upscale module, which is proposed to replace the traditional upscale module. For each scale factor, the SG upscale module provides the corresponding amount of the spatial weights to filter the LR tensor and then converts filtered tensors with the original tensor to corresponding HR images. The proposed method is evaluated through extensive experiments and compared with state-of-the-art (SOTA) methods on widely used benchmark datasets. The experimental results show that our method has superior performance compared with SOTA methods, and the SG upscale module can improve the performance of existing SR networks effectively. What is more, our module has a much less calculation cost than the other upscale modules. [ABSTRACT FROM AUTHOR]

Oxidation of Fe(II) by oxygen (O 2) at circumneutral pH occurs abiotically or is mediated by microaerophilic Fe(II)-oxidizing bacteria. Abiotic Fe(II) oxidation(s) compete with microbial processes and the relative contribution of abiotic reactions depend on the chemical conditions, e.g. P O2 , pH and the presence and identity of ferric (oxyhydr)oxide mineral surfaces, catalyzing the heterogeneous reaction. At circumneutral pH, abiotic Fe(II) oxidation proceeds rapidly, which raises the question how and to which extent neutrophilic microaerophilic Fe(II)-oxidizing bacteria can compete with chemical reactions and gain metabolic energy from microbial Fe(II) oxidation. In this study, we have investigated the environmental constraints for microaerophilic Fe(II) oxidation in a film layer characterized by diffusive supply of both atmospheric O 2 (from the top) and dissolved Fe(II) (from the bottom) by use of a numerical model. A coupled diffusion–reaction model was tested at different chemical (pH and alkalinity gradients) and physical (film layer thickness) parameters to investigate their effects on the relative contributions of different reactions (abiotic homogeneous, heterogeneous and biological Fe(II) oxidation) to the overall (net) Fe(II) oxidation. A first order rate constant for biological oxidation was derived from experimental data from simplifcation of a Monod rate law to be 0.06 h−1. The simulations demonstrate distinct spatial oxidation rate patterns for all of the considered reactions. Microaerophilic Fe(II) oxidation is predominant at a uniform pH 6 and a film thickness z of 1 mm with minor importance at pH 7. Maximum biological rates were on the order of 7∙10-10 mol L−1 s−1 and are in the range of experimentally observed values. Minimum rates were close to the thermodynamic limit. In the presence of a pH gradient and z ≤ 1 mm, two distinct zones were observed: an upper zone dominated by abiotic Fe(II) oxidation (pH ∼ 7) and a lower zone dominated by microaerophilic Fe(II) oxidation (pH < 6.3), while the position and extent of the zones depend on the alkalinity. Such separation is strongly amplified for thinner films (z = 0.2 mm). The importance of heterogenous oxidation depends both on the pH and the amount of ferric (oxyhydr)oxides formed which increases with decreasing diffusive O 2 supply at z > 1 mm. In combination with high resolution imaging of pH values in a biofilm, our simulations underpin the importance of pH gradients in allowing for the formation of microniches. The conditions suitable for microaerophilic Fe(II) oxidation can be predicted based on reaction time scales and three factors were identified to be especially important: i) a pH low enough (<6.3) to outcompete abiotic processes; ii) sufficiently fast diffusive supply of Fe(II) (e.g. by Fe(III)-reducing bacteria or chemical processes) with O 2 concentrations below 150 μmol L−1; iii) sufficient energy gain from Fe(II) oxidation reaction considering the thermodynamic factor F T. We end by discussing strategies that Fe(II)-oxidizing microorganisms can employ to enhance their competitiveness against abiotic reactions. [ABSTRACT FROM AUTHOR]

Change detection in remote sensing images, especially optical high and very high resolution images, is a pivotal technique that enables efficient identification of Earth observation alterations, with widespread applications including land use analysis, urban planning, environmental monitoring, and disaster mapping. Of these applications, change detection is predominantly employed in urbanization studies to monitor developmental changes in construction land. However, the pronounced class imbalance and variances between image domains present formidable challenges to change detection implementations. In response to these issues, we introduce the concept of reverse change discovery, redefining the primary goal of model learning for change detection in optical remote sensing images as stable invariant region detection. In line with this concept, we present a novel Multi-Stage Progressive Change Detection Network (MSP-CD) specifically designed for urbanization change detection, which integrates invariant detection, knowledge distillation, and a coarse-to-fine change detection structure. To validate the effectiveness and robustness of the proposed method, we conducted experiments using two widely recognized public datasets, namely the Lebedev-CD and Levir-CD datasets. The MSP-CD network demonstrated state-of-the-art performance across these datasets. Additionally, we curated an application-specific change detection dataset (JS-CD dataset) using remote sensing images from Jiangsu Province in China, which differs from public datasets in terms of annotations, to further validate the MSP-CD network's performance. A case study on regional urbanization development in Yangzhou city, conducted using the proposed MSP-CD network, further attests to the applicability of our proposed change detection framework. [ABSTRACT FROM AUTHOR]

Osteoporosis (OP), which largely increases the risk of fractures, is the most common chronic degenerative orthopedic disease in the elderly due to the imbalance of bone homeostasis. Alpha-ketoglutaric acid (AKG), an endogenous metabolic intermediate involved in osteogenesis, plays critical roles in osteogenic differentiation and mineralization and the inhibition of osteoclastogenic differentiation. However, the low bioavailability and poor bone-targeting efficiency of AKG seriously limit its efficacy in OP treatment. In this work, a bone-targeting, near-infrared emissive lanthanide luminescence nanocarrier loaded with AKG (β-NaYF 4 :7%Yb, 60%Nd@NaLuF 4 @mSiO 2 -EDTA-AKG, abbreviated as LMEK) is developed for the enhancement of AKG efficacy in OP therapy. By utilizing the NIR-II luminescence (>1000 nm) of LMEK, whole-body bone imaging with high spatial resolution is achieved to confirm the bone enrichment of AKG noninvasively in vivo. The results reveal that LMEK exhibits a remarkable OP therapeutic effect in improving the osseointegration of the surrounding bone in the ovariectomized OP mice models, which is validated by the enhanced inhibition of osteoclast through hypoxia-inducible factor-1α suppression and promotion of osteogenic differentiation in osteoblast. Notably, the dose of AKG in LMEK can be reduced to only 0.2 % of the dose when pure AKG is used in therapy, which dramatically improves the bioavailability of AKG and mitigates the metabolism burden. This work provides a strategy to conquer the low utilization of AKG in OP therapy, which not only overcomes the challenges in AKG efficacy for OP treatment but also offers insights into the development and application of other potential drugs for skeletal diseases. Alpha-ketoglutarate (AKG) is an intermediate within the Krebs cycle, participating in diverse metabolic and cellular processes, showing potential for osteoporosis (OP) therapy. However, AKG's limited bioavailability and inefficient bone-targeting hinder its effectiveness in treating OP. Herein, a near-infrared emissive nanocarrier is developed that precisely targets bones and delivers AKG, bolstering its effectiveness in OP therapy. Thanks to this efficient bone-targeting delivery, the AKG dosage is reduced to 0.2 % of the conventional treatment level. This marks the first utilization of a bone-targeting nanocarrier to amplify AKG's bioavailability and OP therapy efficacy. Furthermore, the mechanism of AKG-loaded nanocarrier regulating the biological behavior of osteoclasts and osteoblasts mediated is tentatively explored. [Display omitted] [ABSTRACT FROM AUTHOR]

Medical Image Super-Resolution plays a pivotal role in enhancing diagnostic accuracy. Transformer-based methods, such as Image Restoration Using Swin Transformer (SwinIR) and Swin transformer for fast Magnetic Resonance Imaging (SwinMR), have shown prowess in this area but also exhibit limitations. Specifically, LayerNorm channel normalization diminishes high-frequency detail, while the Multilayer Perceptron prioritizes global information over local information. Moreover, low-resolution inputs contain substantial low-frequency information, treated uniformly in self-attention, hampering the effectiveness of shifted window-based self-attention. To address these challenges, this study proposes a novel Asymmetric convolution Swin Transformer Layer that leverages global and local information within adjacent windows or pixels. Furthermore, this study presents an innovative joint attention mechanism that integrates channel attention into the Swin Transformer architecture, allowing for the simultaneous capture of local and global information and enhancing the network's ability to adapt and represent information effectively. Based on these components, this study develops a joint Channel Attention and Swin Transformer Residual Network (CSRNet) for medical image super-resolution. To evaluate its performance, this study conducts comprehensive experiments on established datasets consisting of high-resolution medical images. Compared to the state-of-the-art medical image super-resolution methods, the proposed method outperforms in reconstructing medical images across qualitative and quantitative metrics. [ABSTRACT FROM AUTHOR]

The aim of this study was to probe an unexpected relationship between the ice nucleation temperature (T IN), process efficiency and product attributes in a controlled ice nucleation (CIN) lyophilization process. An amorphous product was lyophilized with (CIN-5 °C, CIN-7 °C or CIN-10 °C) or without (NOCIN) control of ice nucleation. Process parameters and product attributes were monitored and compared using a series of advanced in-line and off-line process analytical technology (PAT) tools. Unexpectedly, an indirect relationship was observed between T IN and primary drying efficiency for the CIN processes. Further, the CIN-5 °C process was associated with higher product resistance to mass flow than corresponding CIN-7 °C and CIN-10 °C processes. Surprisingly, the air voids in some NOCIN products were larger than CIN-5 °C products but comparable to CIN-7 °C. Heat flux analysis revealed an indirect relationship between T IN and the minimum hold time required to complete solidification. The heat flux analysis also revealed all products underwent complete solidification prior to primary drying. The order of homogeneity in water activity of the products was CIN-5 °C ≥NOCIN>CIN-7 °C. The higher homogeneity in water activity of CIN-5 °C than corresponding CIN-7 °C processes indicated that the lower process efficiency of CIN-5 °C could not be attributed to unsuccessful induction of ice nucleation during CIN-5 °C. High resolution micro-CT imaging and Artificial Intelligence Image analysis revealed cake wall deformation in CIN-7 °C and NOCIN products but not in CIN-5 °C. In addition, NOCIN products had bimodal distribution in air voids with median size range of 4–5 µm and 151.9–309 µm, respectively, hence the lower process efficiency of NOCIN despite the higher D90. Thus, the observed relationship between T IN and process efficiency may be attributed to microstructural changes post freezing. This hypothesis was corroborated by visible macroscopic cake collapse in NOCIN products but not in CIN products after lyophilization at a higher shelf temperature. In conclusion, the advantages of controlling the ice nucleation temperature of a lyophilization process may only be attained through a robust process design that takes into consideration the primary and secondary drying process parameters. Further, combined use of advanced in-line and off-line PAT tools for process and product characterization may hasten the at scale adoption of advance techniques such as CIN. [ABSTRACT FROM AUTHOR]

In remote sensing image classification, active learning aims to obtain an excellent classification model by selecting informative or representative training samples. However, due to the complexity of remote sensing images, the same class of ground objects usually have different spectral representations. The existing active learning methods may not take into account diverse representations of the same targets, which leads to a possible lack of intra-class diversity in the collected samples. To alleviate this problem, we propose an active learning method based on similarity level histogram (SLH) and adaptive-scale sampling to improve very high resolution remote sensing image classification. Specifically, we construct a SLH for each class of ground objects to effectively consider the intra-class diversity of the same target. To avoid the problem of sample imbalance caused by over-sampling or under-sampling, we design an adaptive-scale sampling strategy. Then, we utilize active learning to mine representative samples from each SLH warehouse according to adaptive-scale sampling strategies until the iteration condition is satisfied. Experiments show that the proposed algorithm can achieve better classification performance with limited training samples and is competitive with other methods based on four sets of publicly available data. • We construct similar-level histograms for each category of object to consider different representations of the same target. • An adaptive-scaling sampling strategy is proposed to avoid sampling unbalance. • We apply active learning and iterative algorithm to mine representative samples through collaboration and interaction. [ABSTRACT FROM AUTHOR]

Light microscopy is the conventional method used to investigate wood anatomy, identify the wood taxon, and assess its conservation state. It generally requires the mechanical cut of thin sections from a sample to obtain informative images. When dealing with wooden artworks or ancient remains (e.g., archaeological waterlogged wood), it is important to avoid sample destruction. In this work the efficacy of micro-magnetic resonance imaging (μ -MRI) to investigate the anatomy of waterlogged wood is assessed in comparison with light microscopy. Images along the three anatomical directions (transverse, tangential and radial) of six modern wood species and one archaeological specimen of waterlogged wood (from the Neolithic site "La Marmotta") were obtained both by μ -MRI and light microscopy. μ -MRI images were acquired virtually selecting 2D slices along the three wood anatomical directions. A 3D reconstruction was derived from 2D μ -MRI images. Conventional light microscopy histology was obtained by manually cutting thin sections. To the best of our knowledge, this is the first study in which high-resolution MR images and light microscopy images of the three anatomical directions of seven wood species are compared. The non-destructive μ -MRI approach allows to investigate the 2D and 3D topological organization of the whole waterlogged wood sample up to a resolution of 8 μm. Although the optical microscope attains higher image resolutions and remains superior in the observation of wood diagnostic characters, multi-parametric μ -MRI provides physiological investigation complementary to light microscopy, giving information concerning both a single section and the whole volume of the sample. The presented study may represent a starting point for further improvements of μ -MRI techniques applied to the non-destructive investigation of waterlogged wood samples, especially those of interest for cultural heritage. [ABSTRACT FROM AUTHOR]

Phototherapy (including photothermal therapy, PTT; and photodynamic therapy, PDT) has been widely used for cancer treatment, but conventional PTT/PDT show limited therapeutic effects due to the lack of disease recognition ability. The integration of fluorescence imaging with PTT/PDT can reveal tumor locations in a real-time manner, holding great potential in early diagnosis and precision treatment of cancers. However, the traditional fluorescence imaging in the visible and near-infrared-I regions (VIS/NIR-I, 400-900 nm) might be interfered by the scattering and autofluorescence from tissues, leading to a low imaging resolution and high false positive rate. The deeper near-infrared-II (NIR-II, 1000-1700 nm) fluorescence imaging can address these interferences. Combining NIR-II fluorescence imaging with PTT/PDT can significantly improve the accuracy of tumor theranostics and minimize damages to normal tissues. This review summarized recent advances in tumor PTT/PDT and NIR-II fluorophores, especially discussed achievements, challenges and prospects around NIR-II fluorescence imaging-guided PTT/PDT for cancers. [Display omitted] [ABSTRACT FROM AUTHOR]

• The image transformation and the super-resolution are integrated for the first time. • We constructed two datasets of small infrared UAV targets to test the performance. • It is possible to obtain a large and clear image simultaneously by STSRGAN. • Style transformation network was designed to maintain shape information and achieve stable learning. With the development of generative adversarial networks, the super-resolution technique of reconstructing a high-resolution image from a low-resolution has achieved excellent resolution results. However, small, low-resolution images are widespread, such as images taken by a thermal camera or with a lens far from the target. Extremely small target image super-resolution is a challenging problem. The main reason is that the small infrared target has fewer pixels and weaker features. The current optimization methods for the tiny target are mainly based on multi-scale feature fusion or super-resolution enhancement. The low-resolution images characterizing small targets are usually obtained by downsampling with high-resolution images during training, which is different from the style of the tiny target in actual detection applications, resulting in poor resolution. In order to solve the problem, we propose a new resolution network: Style Transformation Super-Resolution Generative Adversarial Network (STSRGAN). It contains two sub-networks: one is style transformation GAN to convert the style of the image, and the other is super-resolution GAN. STSRGAN transforms a blurry infrared small target into a clear target with a distribution similar to the training set. Then the resolution can be increased to get a better enhancement effect. The discriminator distinguishes whether the input comes from the generator or the actual image to assist in generating a better super-resolution image. Meanwhile, we produced an infrared Unmanned Aerial Vehicle (UAV) small target dataset with target pixels below 16 × 16. Our method proves better resolution enhancement of small IR targets and shows superior performance over other methods through experiments. [Display omitted] [ABSTRACT FROM AUTHOR]

Recently, Transformer-based models are taken much focus on solving the task of image super-resolution (SR) due to their ability to achieve better performance. However, these models combined huge computational cost during the computing self-attention mechanism. To solve this problem, we proposed a multi-order gated aggregation super-resolution network (MogaSRN) for low-level vision based on the concept of the MogaNet that is developed for high-level vision. The concept of the MogaSRN model is based on spatial multi-order context aggregation and adaptive channel-wise reallocation with the aid of the multi-layer perceptron (MLP). In contrast to the MogaNet model, in which the resolution of each stage decreased by a factor of 2, the resolution of the MogaSRN is stayed fixed during the deep features extraction. Moreover, the structure of the MogaSRN model is built based on balancing the performance and the model complexity. We evaluated our model based on five benchmark datasets concluding that the MogaSRN model can achieve significant improvements compared to the state-of-the-art. Moreover, our model shows the good visual quality and accuracy of the reconstruction. Finally, our model has 3.7 × faster runtime at the scale of × 4 compared to LWSwinIR with better performance. • Image super-resolution based on a deep learning model is proposed. • A MLP multi-order gated aggregation block (MMogaB) is designed. • A MogaSRN model based on utilizing the MMogaB is developed. [ABSTRACT FROM AUTHOR]

The deep-learning-based super-resolution (SR) methods require an avalanche of training images. However, they do not adapt model parameters in test time to cope with the novel blur kernel scenarios. Even though the recent meta-learning-based SR techniques adapt trained model parameters leveraging a test image's internal patch recurrence, they need heavy pre-training on an external dataset to initialize. Also, the shallow internal information exploration and failure to amplify the salient edges impacts blurry SR image generation. Besides, model inferencing gets delayed due to a threshold-dependent adaptation phase. In this paper, we present M ulti-scale Optimized A ttention-aware M eta- L earning framework for SR (MAML-SR) to explore the multi-scale hierarchical self-similarity of recurring patches in a test image. Precisely, without any pre-training, we directly meta-train our model with a second-order optimization having the first-order adapted parameters from the intermediate scales, which are again directly supervised with the ground-truth HR images. At each scale, non-local self-similarity is maximized along with the amplification of salient edges using a novel cross-scale spectro-spatial attention learning unit. Also, we drastically reduce the inference delay by putting a metric-dependent constraint on the gradient updates for a test image. We demonstrate our method's superior super-resolving capability over four benchmark SR datasets. • We propose a meta-learning-based method for image super-resolution. • Multi-scale optimized with deep supervision to rectify latent internal information. • Developed cross-scale attention to extract the salient recurring patches. • Proposed second-order loss based on first-order adapted model to gain stability. • New self-adaptive gradient update strategy in meta-testing to reduce adaptation time. [ABSTRACT FROM AUTHOR]

High-resolution (HR) medical images can provide rich details, which are important for discovering subtle lesions to make diagnoses. Convolutional neural networks (CNNs) are widely used in this field, but struggle to model long-range dependencies. Although transformer-based methods have improved in this respect, this method requires large quantities of data. Unfortunately, large quantities of low-resolution (LR) and HR medical image pairs may not always be available. In addition, most medical image superresolution (SR) methods are deterministic, while the degradation in real scenarios is stochastic. To address these problems, we introduce a probabilistic degradation model that combines natural and medical images for training. This design alleviates the problem of insufficient medical image pairs and learns the degradation process of the natural scene. In addition, we propose a new medical image SR model that consists of CNNs and the Swin Transformer structure to excavate both local and global semantic features. Moreover, to reduce computational stress, the spherical locality-sensitive hashing (SLSH) module is employed in the nonlocal attention (NLA) mechanism to form the ENLA module. This design enables the proposed Sparse Swin Transformer (SSFormer) model to generate HR medical images without extensive training images. Experiments on diverse datasets (natural images and medical images) demonstrate that the proposed method is robust and effective, qualitatively and quantitatively outperforming other medical image SR methods. Code is available at https://github.com/codehxj/SSFormer. • We introduce a probabilistic degradation to model the random factors in real scenarios. This model combines natural images and medical images for training, which alleviates insufficient medical images and enables the generated low-resolution medical images to learn the authentic degradation process in natural images. • We proposed a new medical SR model, SSFormer, which integrates transformer components and CNN modules into a united framework to capture the long-range dependencies and local perception. • We use the nonlocal attention mechanism to achieve a similar effect. Meanwhile, to reduce the computational cost, the locality sensitive hashing is employed in the nonlocal attention module by projecting extracted features onto a spherical hyperplane for similarity calculations. [Display omitted] [ABSTRACT FROM AUTHOR]

Multi-modal image fusion is widely used in various fields, especially in military, medical, industrial detection and other fields. Image fusion can integrate redundant and complementary information of two or more multi-modal images into one image, so that the fused image contains more useful information. In this paper, we construct a novel dataset for Multi-modal Image Fusion Applications (MOFA), including four modals: visible, near-infrared (NIR), long-wavelength infrared (LWIR) and polarization. The MOFA dataset contains 1062 images of 118 groups, in which 450 are indoor and 612 are outdoor. The dataset is applied to different image fusion applications, including general multi-modal image fusion, fusion based image super-resolution and image restoration. Multiple image fusion methods are compared and analyzed on this dataset with a qualitative assessment of subjective and objective metrics. Based on the experiments, the advantages and disadvantages of different methods are discussed. Moreover, the challenging problems of image fusion are concluded. • Build a multi-modal imaging device. • Construct a multi-modal dataset, including visible, NIR, LWIR and polarization. • Explore the effectiveness of the dataset for different image processing tasks. • Analyze the advantages and disadvantages of different fusion methods. [ABSTRACT FROM AUTHOR]

Over the past few years, single image super-resolution (SR) has become a hotspot in the remote sensing area, and numerous methods have made remarkable progress in this fundamental task. However, they usually rely on the assumption that images suffer from a fixed known degradation process, e.g., bicubic downsampling. To save us from performance drop when real-world distribution deviates from the naive assumption, blind image super-resolution for multiple and unknown degradations has been explored. Nevertheless, the lack of a real-world dataset and the challenge of reasonable degradation estimation hinder us from moving forward. In this paper, a self-supervised degradation-guided adaptive network is proposed to mitigate the domain gap between simulation and reality. Firstly, the complicated degradations are characterized by robust representations in embedding space, which promote adaptability to the downstream SR network with degradation priors. Specifically, we incorporated contrastive learning to blind remote sensing image SR, which guides the reconstruction process by encouraging the positive representations (relevant information) while punishing the negatives. Besides, an effective dual-wise feature modulation network is proposed for feature adaptation. With the guide of degradation representations, we conduct modulation on feature and channel dimensions to transform the low-resolution features into the desired domain that is suitable for reconstructing high-resolution images. Extensive experiments on three mainstream datasets have demonstrated our superiority against state-of-the-art methods. Our source code can be found at https://github.com/XY-boy/DRSR • A simple and effective self-supervised degradation representation learning strategy is proposed. • The learned representations can promote cross-domain super-resolution generalization. • A dual-wise modulation network is established for effective feature adaption. • Our method achieves favorable performance on various remote sensing datasets. [ABSTRACT FROM AUTHOR]

• An advanced Reference-Constrained Image Restoration algorithm is proposed for solving the image restoration task. • Our algorithm takes advantage of image prior information and merges it with Regularization by Denoising. • By using multiple reference images, we increase the possibility to improve the quality of the reconstructed images. • The achievable performance of the proposed algorithm outperforms Regularization by Denoising. Image restoration is always a long-term problem in low-level computer vision, which has many practical applications. In this paper, we devote to reconstructing a single image from compressive measurements with the assistance of one or multiple reference images, by using the similarity prior information of them. More specifically, we propose to bring in the concept of Regularization by Denoising, which leverages existing denoising engine to regularize inverse problems, and illustrate how the reference images and this powerful algorithm can be merged into a new highly effective recovery approach, which is named Reference Image Constrained Regularization by Denoising for solving various image restoration tasks. Since the similarity between the reference and the ground truth is not guaranteed, or the degree of the similarity is different, we extend the proposed algorithm by adding adaptive weighting matrix to the difference between the target image and the reference images to further improve the performance. We evaluate experimentally the proposed algorithm in the image deblurring and single image super-resolution problems. Furthermore, we also carry out our method on video super-resolution task, in the case of multiple reference images available. We compare the advanced reference-constrained image restoration algorithm with previous methods about reconstruction accuracy and computation cost. The experimental results show that the proposed method outperforms state-of-the-art algorithms, including Regularization by Denoising, in terms of both the Peak Signal to Noise Ratio and the visual performance. [ABSTRACT FROM AUTHOR]

Scanning acoustic microscopy (SAM) technique has been applied to defect inspection in electronic devices. With the increase of packaging density, detection of the micro-defects in high density devices becomes more and more challenging. The SAM test is suffering from sacrificing the spatial resolution to reach a certain penetration depth of the ultrasonic waves. So it is necessary to enhance the resolution level of the SAM image. In this paper, a wavelet based resolution enhancement technique was investigated to reconstruct a high quality image for SAM test of the flip chip packages. The stationary wavelet transform was adopted to decompose the captured SAM image into four frequency subbands, and the high frequency subbands were enhanced by adding the difference matrix in the intermediate stage, and a super resolution SAM image was derived from combining all the subbands by using the Inverse Discrete Wavelet Transform. Then the solder joints segmented from the SR-SAM image were classified by using the SVM algorithm. The results validated that the proposed technique is effective to improve the detection accuracy of SAM test. • An intelligent system is developed to inspect the solder joints of flip chip. • The SAM image is decomposed into four different frequency components. • High resolution SAM image is reconstructed with the wavelet-based method. • The solder joints are segmented from the SR-SAM image. • The SR-SAM image has high accuracy for defect recognition. [ABSTRACT FROM AUTHOR]

The performance of SISR models is currently growing at a rapid pace since the adoption of Convolutional Neural Networks (CNN) in the domain. Despite that, the understanding of the underlying factors of this rapid growth is still low. In this study, we proposed a study design to measure the impact of the inductive bias in CNN on its training performance, which is one of the factors of the excellent performance of CNN in SISR. With this study design, we observed that not all CNN has a supportive inductive bias to the training performance. This result can provide guidance to future research in choosing a suitable CNN architecture for optimal SISR performance. [ABSTRACT FROM AUTHOR]

To understand the process of mobilisation of shear strength in sand-geosynthetic interfaces at a fundamental level, it is essential to precisely characterize the size and shape of the grains and the shear-induced surface changes in geosynthetics. In the current study, shear behaviour of dilative and non-dilative geosynthetics interfacing with sands of different morphological characteristics was analysed through interface shear tests and a gamut of digital imaging techniques. 3D shape parameters of sands such as sphericity, convexity, roundness, aspect ratio, and roughness were quantified at different scales using X-ray micro computed tomography (μCT) and optical profilometry. Interface shear tests revealed higher peak and residual friction angles for particles with greater irregularity, angularity, and surface texture. The surface texture of geotextile surfaces resulted in higher interface friction and higher vertical displacement compared to geomembrane surfaces, which showed completely non-dilative behaviour. Surface changes in geomembranes were quantified using laser profilometry. High resolution images obtained at different stages of shearing were analysed for quantifying the shear zone thickness using digital image correlation (DIC). Thickness of the shear bands, microscopic shearing mechanisms and shear strength are correlated to the multi-scale shape parameters of sands and surface changes in geosynthetic surfaces. • Shear tests are carried out on sands, sand-geotextile and sand-geomembrane interfaces. • Sand particle shape is characterized using μ-CT scanning and spherical harmonic particle reconstruction. • Shear zone thickness is measured using DIC correlations. • Surface changes to geomembrane are quantified using laser profilometry. • Particle shape effects on interface shear strength is quantified. [ABSTRACT FROM AUTHOR]

Deep neural networks can significantly improve the quality of super-resolution. However, previous work has made insufficient use of low-resolution scale features and channel-wise information, hence hindering the representational ability of CNNs. To address these issues, a multi-scale fractal residual attention network (MFRAN) is proposed. Specifically, MFRAN consists of fractal residual blocks (FRBs), dual-enhanced channel attention (DECA), and dilated residual attention blocks (DRABs). Among them, FRB applies multi-scale extension rule to continuously expand into a fractal structure that detects multi-scale features; DRAB constructs a combined dilated convolution to learn a generalizable and expressive feature space with a larger receptive field; DECA employs one-dimensional convolution to achieve cross-channel information interaction, and enhance the flow of information between groups by channel shuffling. Then, we integrate horizontal feature representations via local residual and feature fusion. Extensive quantitative and qualitative evaluations of benchmark datasets show that our proposed approach outperforms state-of-the-art methods in terms of quantitative metrics and visual results. [Display omitted] • We propose a Multi-scale Fractal Residual Attention Network (MFRAN) for SISR, where Dual-Enhanced Channel Attention can capture inter-channel dependencies more efficiently and lightly while enabling cross-channel information interaction for enhanced channel modeling capabilities, so that the model can reconstruct SR images with richer details at large-scale factors; Fractal Residual Block with multiple local feature extraction paths of different lengths, each path consists of dilated residual attention blocks with different size receptive fields, so that LR features of different scales can be extracted efficiently. [ABSTRACT FROM AUTHOR]

The drying time of lyophilization and resultant cake microstructure are dependent on each other as water and solvent leave a lyophilized cake. The drying rate affects the size, distribution, and tortuosity of the pores as these macropores evolve during the primary drying phase, which in return impact the further removal of water and solvent from the cake throughout the drying period. This interplay results in a microstructure that determines the reconstitution time for a given formulation. The current study employs advanced X-ray Microscopy (XRM) coupled with mathematical models to correlate the microstructure with the drying kinetics and the reconstitution time. The normalized diffusion coefficients, derived from the reconstructed 3D microstructure of the cake, correlate with the solid content of the pre-lyophilization solution and agree with the mass transfer coefficients from a semi-empirical drying model built with lyophilization process data. Specifically, a solution with less solid content leads to a lyophilized cake with larger pores, thinner walls, and a greater pore volume compared to a solution with more solid content. Consequently, models from the microstructure and drying experiments reveals faster mass transfer independently. While the mass transfer models from the cake structure and the lyophilization process data accurately represents the drying kinetics, both models are inadequate to describe the reconstitution process due to the significant impact from formulation ingredients that alter the mass transfer mechanism via solubility and wettability. In summary, X-ray microscopy imaging and mathematical models are powerful tools that provide insights into the lyophilization process from a new angle. [Display omitted] • Pore characteristics, such as diameter (distribution), wall thickness (distribution), and pore volume were quantitative determined by high resolution XRM images. • The quantification of pore characteristics benefited from machine learning algorithms that handled the large number of images collected in this study. • Two models were used to describe the drying process: 1) A transportation model based on reconstructed 3D pore images was built and used to estimated drying time from effective diffusivity with finite volume method simulation; 2) an experimental data based drying model that derived mass transfer resistance. The two independent models agreed well when describing the drying process • Formulation elements had a great impact on the reconstitution time where micropore structure alone was insufficient to explain the different reconstitution behavior. [ABSTRACT FROM AUTHOR]

The Cdv system is a protein machinery responsible for cell division in members of the TACK superphylum of archaea. The Cdv system presents many structural similarities to the eukaryotic ESCRT machinery, and it is presumed to act in a similar manner. The Cdv system has been found to be implied in a wide range of processes in the cells, such as division, vesicle budding, and virus release. The role of the CdvB paralogs is key in the division process, as became clear through recent developments in high-temperature live-cell imaging and super-resolution microscopy. In vitro experiments have shed light on the mutual interactions between CdvA, CdvC, and the various CdvB paralogs. The Cdv system is the protein machinery that performs cell division and other membrane-deforming processes in a subset of archaea. Evolutionarily, the system is closely related to the eukaryotic ESCRT machinery, with which it shares many structural and functional similarities. Since its first description 15 years ago, the understanding of the Cdv system progressed rather slowly, but recent discoveries sparked renewed interest and insights. The emerging physical picture appears to be that CdvA acts as a membrane anchor, CdvB as a scaffold that localizes division to the mid-cell position, CdvB1 and CvdB2 as the actual constriction machinery, and CdvC as the ATPase that detaches Cdv proteins from the membrane. This paper provides a comprehensive overview of the research done on Cdv and explains how this relatively understudied machinery acts to perform its cell-division function. Understanding of the Cdv system helps to better grasp the biophysics and evolution of archaea, and furthermore provides new opportunities for the bottom-up building of a divisome for synthetic cells. [ABSTRACT FROM AUTHOR]

The characteristics of synthetic aperture radar (SAR) images are easily affected by factors such as sensor parameters, imaging scenes, etc., which may lead to data distributional discrepancies between the ideal training phase and the practical application phase, resulting in generalization performance degradation of deep learning based detectors in practice. In this paper, an accurate and fast domain adaptive detection method is proposed to address the performance degradation problem of cross-scene detection without requiring additional manual annotations. The cross-scene detection performance is gradually improved by confidence-induced unsupervised feature adaptation (CUFA) and uncertainty-aware adaptive pseudo-label learning (UAPL). In CUFA, image-level and instance-level feature adaptation are progressively implemented to bridge the distribution discrepancies between the source and the target domain in the latent feature space. Then, UAPL is performed to adaptively introduce accurate and stable supervisions of target domain according to the learning status of the detector, avoiding error accumulations and further improving the cross-scene target detection performance. The proposed cross-scene detection method is validated on airborne and spaceborne SAR image datasets of different imaging scenes. The experimental results show that by combining both CUFA and UAPL, compared with the second-best comparison method, the proposed method significantly improves the cross-scene target detection performance in SAR images (at least + 6.5% AP50 and F 1 in miniSAR and FARAD cross-scene tasks, and at least + 7.5% AP50 and F 1 in Gaofen-3 and Sentinel-1 cross-scene tasks), which facilitates application of the deep learning based detectors in the realistic scenarios. [ABSTRACT FROM AUTHOR]

Recently, super-resolution (SR) techniques based on deep learning have taken more and more attention, aiming to improve the images and videos resolutions. Most of the SR methods are related to other fields of computer vision such as image classification, image segmentation, and object detection. Based on the success of the image SR task, many image SR surveys are introduced to summarize the recent work in the image SR domains. However, there is no survey to summarize the SR models for the lightweight image SR domain. In this paper, we present a comprehensive survey of the state-of-the-art lightweight SR models based on deep learning. The SR techniques are grouped into six major categories: include convolution, residual, dense, distillation, attention, and extremely lightweight based models. Also, we cover some other issues related to the SR task, such as benchmark datasets and metrics for performance evaluation. Finally, we discuss some future directions and open problems, that may help other community researchers in the future. [Display omitted] • Lightweight image super-resolution based on deep learning models are reviewed. • The network design, loss function, framework, and training datasets are considered. • The network design structure of the reviewed models are categorized into six types. • The future research directions of lightweight image super-resolution are indicated. [ABSTRACT FROM AUTHOR]

• We combine the domian knowledge of HSI with deep learning. • We develop an efficient unfolding network (EUNet) for HSI SR. • We design a lightweight HSI SR prior network with disentangled spatial-spectral representation. • Our EUNet can achieve comparable results in different scenarios with limited computing resources. Hyperspectral image super-resolution (HSI SR) is dramatically impacted by high spectral dimensionality, insufficient spatial resolution, and limited availability of training samples. Current approaches mainly rely on complex data-driven models to address some of these challenges, and the characteristics of HSI are not fully considered in the model design. In this paper, we propose an efficient unfolding network with disentangled spatial-spectral representation (EUNet) for HSI SR by combining domain knowledge (i.e., spectral correlation, degradation model, and structure prior) with deep learning. Specifically, the optimization process of the super-resolution prior-driven Maximum A Posterior (MAP) framework is unfolded into an interpretable multi-stage network, which inherits the advantages of deep learning-based image super-resolution (e.g., feature extraction in low-resolution space) and explicitly imposes the degradation model constraint. To well incorporate the structure prior of HSI, spatial and spectral feature extraction is disentangled by a variant of depthwise separable convolution, and spectral correlation is embedded by a lightweight spectral attention mechanism, so that the difficulty and computational complexity of feature learning are greatly reduced. Experiments on benchmark datasets with different degradation models demonstrate the feasibility and superiority of the proposed EUNet over other state-of-the-art methods in terms of evaluation metrics and computational complexity. The source code is available at https://github.com/denghong-liu/EUNet. [ABSTRACT FROM AUTHOR]

Hepatocellular carcinoma (HCC) is a highly prevalent form of liver cancer diagnosed annually in 600,000 people worldwide. A common treatment is transarterial chemoembolization (TACE), which interrupts the blood supply of oxygen and nutrients to the tumor mass. The need for repeat TACE treatments may be assessed in the weeks after therapy with contrast-enhanced ultrasound (CEUS) imaging. Although the spatial resolution of traditional CEUS has been restricted by the diffraction limit of ultrasound (US), this physical barrier has been overcome by a recent innovation known as super-resolution US (SRUS) imaging. In short, SRUS enhances the visible details of smaller microvascular structures on the 10 to 100 µm scale, which unlocks a host of new clinical opportunities for US. In this study, a rat model of orthotopic HCC is introduced and TACE treatment response (to a doxorubicin-lipiodol emulsion) is assessed using longitudinal SRUS and magnetic resonance imaging (MRI) performed at 0, 7 and 14 d. Animals were euthanized at 14 d for histological analysis of excised tumor tissue and determination of TACE response, that is, control, partial response or complete response. CEUS imaging was performed using a pre-clinical US system (Vevo 3100, FUJIFILM VisualSonics Inc.) equipped with an MX201 linear array transducer. After administration of a microbubble contrast agent (Definity, Lantheus Medical Imaging), a series of CEUS images were collected at each tissue cross-section as the transducer was mechanically stepped at 100 μm increments. SRUS images were formed at each spatial position, and a microvascular density metric was calculated. Microscale computed tomography (microCT, OI/CT, MILabs) was used to confirm TACE procedure success, and tumor size was monitored using a small animal MRI system (BioSpec 3T, Bruker Corp.). Although there were no differences at baseline (p > 0.15), both microvascular density levels and tumor size measures from the complete responder cases at 14 d were considerably lower and smaller, respectively, than those in the partial responder or control group animals. Histological analysis revealed tumor-to-necrosis levels of 8.4%, 51.1% and 100%, for the control, partial responder and complete responder groups, respectively (p < 0.005). SRUS imaging is a promising modality for assessing early changes in microvascular networks in response to tissue perfusion-altering interventions such as TACE treatment of HCC. [ABSTRACT FROM AUTHOR]

Aberrant patterns of cognition, perception, and behavior seen in psychiatric disorders are thought to be driven by a complex interplay of neural processes that evolve at a rapid temporal scale. Understanding these dynamic processes in vivo in humans has been hampered by a trade-off between spatial and temporal resolutions inherent to current neuroimaging technology. A recent trend in psychiatric research has been the use of high temporal resolution imaging, particularly magnetoencephalography, often in conjunction with sophisticated machine learning decoding techniques. Developments here promise novel insights into the spatiotemporal dynamics of cognitive phenomena, including domains relevant to psychiatric illnesses such as reward and avoidance learning, memory, and planning. This review considers recent advances afforded by exploiting this increased spatiotemporal precision, with specific reference to applications that seek to drive a mechanistic understanding of psychopathology and the realization of preclinical translation. [ABSTRACT FROM AUTHOR]

The availability of high-resolution data and accurate ground truth is essential to evaluate and compare methods and algorithms properly. Moreover, it is often difficult to acquire real data for a given application domain that is sufficiently representative and heterogeneous in terms of scene representation, acquisition conditions, point of view, etc. To overcome the limitations of available datasets, this paper presents a new synthetic, multi-purpose dataset called ENRICH for testing photogrammetric and computer vision algorithms. Compared to existing datasets, ENRICH offers higher resolution images rendered with different lighting conditions, camera orientations, scales, and fields of view. Specifically, ENRICH is composed of three sub-datasets: ENRICH-Aerial, ENRICH-Square, and ENRICH-Statue, each exhibiting different characteristics. We show the usefulness of the proposed dataset on several examples of photogrammetry and computer vision-related tasks such as: evaluation of hand-crafted and deep learning-based local features, effects of ground control points (GCPs) configuration on the 3D accuracy, and monocular depth estimation. We make ENRICH publicly available at: https://github.com/davidemarelli/ENRICH. [ABSTRACT FROM AUTHOR]

Detection and conversion of mechanical forces into biochemical signals is known as mechanotransduction. From cells to tissues, mechanotransduction regulates migration, proliferation, and differentiation in processes such as immune responses, development, and cancer progression. Mechanosensitive structures such as integrin adhesions, the actin cortex, ion channels, caveolae, and the nucleus sense and transmit forces. In vitro approaches showed that mechanosensing is based on force-dependent protein deformations and reorganizations. However, the mechanisms in cells remained unclear since cell imaging techniques lacked molecular resolution. Thanks to recent developments in super-resolution microscopy (SRM) and molecular force sensors, it is possible to obtain molecular insight of mechanosensing in live cells. We discuss how understanding of molecular mechanotransduction was revolutionized by these innovative approaches, focusing on integrin adhesions, actin structures, and the plasma membrane. Mechanotransduction is involved in a variety of cell and tissue processes. For this, mechanosensitive structures, such as integrin adhesions, the cytoskeleton, or the plasma membrane, are key to sense, integrate, and transmit mechanical forces. Novel imaging methods and molecular sensors have unveiled how molecular mechanosensing occurs in these structures within the context of live cells. These include super-resolution imaging, single particle tracking, and molecular force sensors. Super-resolution microscopy and single protein tracking have revealed the 3D nanoscale organization and dynamics of integrin adhesion structures in various cell types, which are divided into functional nanolayers with specific paths for protein diffusion. Forces on actin regulators and F-actin trigger conformational changes controlling actin regulator function and binding, and therefore F-actin assembly and architecture. In migrating cells, mechanical plasticity emerges from global geometrical reorganization of actin networks under loads, or forces from elongating actin filaments controlling locally actin regulators dynamics and functions. Cell surface mechanics are emerging as a key parameter controlling cell motility, division, and differentiation. Variations in cortical tension and membrane-to-cortex attachment regulate the timing of differentiation while the nanoscale architecture of the actin cortex is linked to cortical tension. [ABSTRACT FROM AUTHOR]

• Detailed survey on recent advancements in image super resolution. • Broad Classification of methods in three categories. • Benchmark algorithms are discussed with experimental analysis. Super resolution (SR) is an eminent system in the field of computer vison and image processing to improve the visual perception of the poor-quality images. The key objective of image super resolution is to address the limitations of imaging systems mainly due to hardware problems and requirements for clinical processing of medical imaging using post-processing operations. Numerous super resolution strategies have been put-forward in the computer vision community to improve and achieve high-resolution images over the years. In the past few years, there has been a significant advancement in image super-resolution algorithms. This paper aims to provide the detailed survey on recent advancements in image super-resolution in terms of traditional, deep learning and the latest transformer-based algorithms. The in-depth taxonomy of broadly classified super-resolution techniques within these categories has been broadly discussed. An extensive survey has been carried out on deep learning techniques in terms of parameters, architecture, network complexity, depth, learning rate, framework, optimization, and loss function. Furthermore, we also address some of the significant parameters such as problem definition, evaluation metrics, publicly benchmarks datasets, loss functions and applications. In addition, we have performed an experimental analysis and comparison of various benchmark algorithms on publicly available datasets both qualitively and quantitively. Lastly, we conclude our survey by emphasizing some of the prospective future directions and open issues that the community need to address in the future. [ABSTRACT FROM AUTHOR]

• We proposed a jointly texture enhanced and stereo captured attention network for stereo image super-resolution. • We proposed a joint texture and parallax attention module (JTPAM) for texture enhancement and stereo capturing. • The texture attention module in JTPAM is no-reference based. • Our network achieves the state-of-the-art (SOTA) performance over the compared methods in some authoritative databases. Stereo image super-resolution (StereoISR) aims at recovering high resolution (HR) details and realistic textures from low-resolution (LR) left-right pairs with stereo correspondence. Recently, some advanced 2D methods have achieved good performance in realistic texture reconstruction via texture attention mechanism in a reference-based training way. However, the existing 3D super-resolution (SR) datasets lack the HR reference to obtain the LR-to-HR reference pair for the reference-based training. To overcome this problem, we propose a novel StereoISR network with a joint texture and parallax attention module (JTPAM), where the down-sampled LR (named super-low-resolution (SLR)) image and the LR image are taken as the SLR-to-LR reference pair, to simulate the texture conversion relationship between the LR and HR images. Additionally, the parallax attention mechanism is utilized to maintain the stereo correspondence by calculating the feature similarity along the Epipolar line. Experiment results on the Flickr1024, Middlebury, KITTI 2012 and KITTI 2015 datasets show that our network achieves the state-of-the-art (SOTA) performance over the compared methods, and the stereo SR images generated by our network have abundant binocular information and clear texture details. [Display omitted] [ABSTRACT FROM AUTHOR]

Advanced technology make more fine clarity image in US I would want to introduce high resolution ultrasound imae in superficial lesion make by high Mhz (18Mhz hockey sticky probe,15MHZ matrix probe and 8 to12 MHZ linear)probe and supporting machine. I would want to show clinically usefulness of high resolution ultrasound image and different from low clarity images. High resolution US show more detail finding in ligament and sheath disease. such as De Quervain disease figure out sheath hypervascularity by PD, Rotator interval consist structure injury, superfial ligament injury and intra ligament calcium thread formation. High resolution US have capabilty of not only delineated sophiscated anatomical structure of wrist,hand, finger but also figure out small lesion too difficult to evaluation on LOW frequency probe such as FCR tendon sheath tear point, peroneus longus te ndon small tear and effusion in peroneal retinaculum and ECU complex sheath tear and tendon substance gradation injury. For maximised high quality US image, we need high frequency probe such as high Mhz hockey stick, good software for clear image processing,upgrade main body for fast response, 4k rapid and non-flick monitor and operator's corrective set up for imag e purpose. High resolution (15-18Mhz) US are supplied more clear image about small and delicate lesion, and helpful for clinical decision making,patient evaluation and widely useful for various field MSK disease. High resolution US is clear and sensitive than MR in limited superficial structure and ligament image. High resolution (15-18Mhz) US are supplied more clear image about small and delicate lesion, and helpful for clinical decision making,patient evaluation and widely useful for various field MSK disease. High resolution US is clear and sensitive than MR in limited superficial structure and ligament image. [ABSTRACT FROM AUTHOR]

In single image super-resolution (SISR), deep neural networks learn mainly nonlinear functions to obtain promising high-resolution (HR) images. However, there are usually two undesired limitations to recovered images of existing SR methods. First, since this task is typically an ill-posed problem, recovering the structural information for the SR process is usually sharp edges but distorted. Second, since this task generally has an extremely large space for the mapping function, learning this mapping from low-resolution (LR) to HR images is typically difficult. For the most part, SR models usually suffer from the lack of structural information for the objects in images and poor performance. To address the above issues, we propose a dual gradient regression scheme in the first stage of the network to recover the structure information of objects in images and the corresponding loss function to learn the difference between structural maps. In the second stage of the network, we restore HR gradient maps by the first stage to provide additional structural information for the second stage, further constraining the structure of the SR image and reducing the space of the possible functions. Extensive experiments demonstrate our results outperform the state-of-the-art SR methods in the SSIM, and achieve better visual effects than the state-of-the-art SR methods. [Display omitted] • In this manuscript, we propose a dual gradient regression scheme and corresponding loss function for image super-resolution. • First, since this task is typically an ill-posed problem, recovering the structural information for the SR process is usually sharp edges but distorted. • Second, since this task generally has an extremely large space for the mapping function, learning this mapping from low-resolution (LR) to HR images is typically difficult. • To address the above issues, we propose a dual gradient regression scheme in the first stage of the network to recover the structure information of objects in images and the corresponding loss function to learn the difference between structural maps. • In the second stage of the network, we restore HR gradient maps by the first stage to provide additional structural information for the second stage, further constraining the structure of the SR image and reducing the space of the possible functions. • Extensive experiments substantiate the superiority of the proposed network and prove that our method outperforms the current state-of-the-art method. [ABSTRACT FROM AUTHOR]