Search

In July 2021, numerous audio recordings dating back to 2006 through 2012 were leaked. In them, Real Madrid's President Florentino Pérez was heard insulting some of the most beloved legends of Real Madrid fans, including celebrity Cristiano Ronaldo. Despite his lack of manners, the seriousness of the president's insults, and the lack of a formal apology, the images of Pérez and Real Madrid barely suffered because of this incident. This scholarly commentary examines the variables of culture and country as well as the specific characteristics of the soccer industry and fandom that explain the lack of public accountability when a top executive of a soccer organization commits verbal excesses in form and substance, as in the Pérez case. [ABSTRACT FROM AUTHOR]

In this paper, due to factors such as imaging devices and imaging environment, medical images may experience halos, blurring, and unclear contrast. A medical image restoration method based on neighborhood registration is proposed to address this issue. In this repair method, the image is first processed into blocks, then the blocks that need to be repaired are determined. Afterward, based on the surrounding neighborhood information of the image block to be repaired, the image information within the optimal neighborhood block is repaired and replaced. The experimental results show that the repair method based on the best neighborhood registration has completed medical image restoration, with significantly higher repair quality than the reference method. The repaired image results have a high RSNR. [ABSTRACT FROM AUTHOR]

In response to the inadequate utilization of prior information in current mural inpainting processes, leading to issues such as semantically unreliable inpaintings and the presence of artifacts in the inpainting area, a Dunhuang mural inpainting method based on reference guidance and multi‐scale feature fusion is proposed. First, the simulated broken mural, the mask image, and the reference mural are input into the model to complete the multi‐level embedding of patches and align the multi‐scale fine‐grained features of damaged murals and reference murals. Following the patch embedding module, a hybrid residual module is added based on hybrid attention to fully extract mural features. In addition, by continuing the residual concatenation of outputs of the hierarchical embedding module improves the ability of the model to represent deeper features, and improves the robustness and generalisation of the model. Second, the encoded features are fed into the decoder to generate decoded features. Finally, the convolutional tail is employed to propagate them and complete the mural painting. Experimental validation on the Dunhuang mural dataset demonstrates that, compared to other algorithms, this model exhibits higher evaluation metrics in the inpainting of extensively damaged murals and demonstrates overall robustness. In terms of visual effects, the results of this model in the inpainting process exhibit finer textures, richer semantic information, more coherent edge structures, and a closer resemblance to authentic murals. [ABSTRACT FROM AUTHOR]

Methods based on the physical Retinex model are effective in enhancing low-light images, adeptly handling the challenges posed by low signal-to-noise ratios and high noise in images captured under weak lighting conditions. However, traditional models based on manually designed Retinex priors do not adapt well to complex and varying degradation environments. DEANet (Jiang et al., Tsinghua Sci Technol. 2023;28(4):743–53 2023) combines frequency and Retinex to address the interference of high-frequency noise in low-light image restoration. Nonetheless, low-frequency noise still significantly impacts the restoration of low-light images. To overcome this issue, this paper integrates the physical Retinex model with deep learning to propose a joint network model, DEANet++, for enhancing low-light images. The model is divided into three modules: decomposition, enhancement, and adjustment. The decomposition module employs a data-driven approach based on Retinex theory to split the image; the enhancement module restores degradation and adjusts brightness in the decomposed images; and the adjustment module restores details and adjusts complex features in the enhanced images. Trained on the publicly available LOL dataset, DEANet++ not only surpasses the control group in both visual and quantitative aspects but also achieves superior results compared to other Retinex-based enhancement methods. Ablation studies and additional experiments highlight the importance of each component in this method. [ABSTRACT FROM AUTHOR]

In computer vision, eliminating the effects of adverse weather conditions such as rain, snow, and fog on images is a key research challenge. Existing studies primarily focus on image restoration for single weather types, while methods addressing image restoration under multiple combined weather conditions remain relatively scarce. Furthermore, current mainstream restoration networks, mostly based on Transformer and CNN architectures, struggle to achieve an effective balance between global receptive field and computational efficiency, limiting their performance in practical applications. This study proposes ACMamba, an end-to-end lightweight network based on selective state space models, aimed at achieving image restoration under multiple weather conditions using a unified set of parameters. Specifically, we design a novel Visual State Space Module (VSSM) and a Spatially Aware Feed-Forward Network (SAFN), which organically combine the local feature extraction capabilities of convolutions with the long-range dependency modeling capabilities of selective state space models (SSMs). This combination significantly improves computational efficiency while maintaining a global receptive field, enabling effective application of the Mamba architecture to multi-weather image restoration tasks. Comprehensive experiments demonstrate that our proposed approach significantly outperforms existing methods for both specific and multi-weather tasks across multiple benchmark datasets, showcasing its efficient long-range modeling potential in multi-weather image restoration tasks. [ABSTRACT FROM AUTHOR]

Face shape priors such as landmarks, heatmaps, and parsing maps are widely used to improve face super resolution (SR). It is observed that face priors provide locations of high‐frequency details in key facial areas such as the eyes and mouth. However, existing methods fail to effectively exploit the high‐frequency information by using the priors as either constraints or inputs. This paper proposes a novel high frequency highway (H2F${\rm H}_2{\rm F}$) framework to better utilize prior information for face SR, which dynamically decomposes the final SR face into a coarse SR face and a high frequency (HF) face. The coarse SR face is reconstructed from a low‐resolution face via a texture branch, using only pixel‐wise reconstruction loss. Meanwhile, the HF face is directly generated from face priors via an HF branch that employs the proposed inception–hourglass model. As a result, H2F${\rm H}_2{\rm F}$ allows the face priors to have a direct impact on the SR face by adding the outputs of both branches as the final result and provides an extra face editing function. Extensive experiments show that H2F${\rm H}_2{\rm F}$ significantly outperforms state‐of‐the‐art face SR methods, is general for different texture branch models and face priors, and is robust to dataset mismatch and pose variations. [ABSTRACT FROM AUTHOR]

Image super‐resolution (SR) has been widely applied in remote sensing to generate high‐resolution (HR) images without increasing hardware costs. However, SR is a severe ill‐posed problem. As deep learning advances, existing methods have solved this problem to a certain extent. However, the complex spatial distribution of remote sensing images still poses a challenge in effectively extracting abundant high‐frequency details from the images. Here, a single‐image super‐resolution (SISR) network based on the generative adversarial network (GAN) for remote sensing is presented, called JOA‐GAN. Firstly, a joint‐attention module (JOA) is proposed to focus the network on high‐frequency regions in remote sensing images to enhance the quality of image reconstruction. In the generator network, a multi‐scale densely connected feature extraction block (ERRDB) is proposed, which acquires features at different scales using MSconv blocks containing multi‐scale convolutions and automatically adjusts the features by JOA. In the discriminator network, the relative discriminator is used to compute the relative probability instead of the absolute probability, which helps the network learn clearer and more realistic texture details. JOA‐GAN is compared with other advanced methods, and the results demonstrate that JOA‐GAN has improved objective evaluation metrics and achieved superior visual effects. [ABSTRACT FROM AUTHOR]

The tensor singular value decomposition‐based model has garnered increasing attention in addressing tensor recovery challenges. However, existing tensor recovery methods exhibit certain inherent limitations. Some ignore the simultaneous effects of noise and missing values, while most can't handle higher‐order tensors, which are not reflective of real‐world scenarios. The information redundancy within tensor data often leads to a prevailing low‐rank structure, making low‐rankness a vital prior in the tensor recovery process. To tackle this pressing issue, a robust low‐rank tensor recovery framework is proposed to rehabilitate higher‐order tensors corrupted by sparse noise and missing entries. In the model, the tensor nuclear norm derived for order‐d tensors (d≥$\ge$ 4) are employed as a representation of the low‐rank prior, while utilizing the L1$\mathtt {L}_1$‐norm to model the sparse noise. To solve the proposed model, an efficient Alternating direction method of multipliers algorithm is developed. A series of experiments are performed on synthetic and real‐world datasets. The results show that the superior performance of the method compared with other algorithms dedicated to addressing order‐d tensor recovery challenges. Notably, in scenarios where the data is severely compromised (noise ratio 40%, sample ratio 70%), the algorithm consistently outperforms its competitors, achieving significantly improved results. [ABSTRACT FROM AUTHOR]

This study investigates penumbrae and light bridges based on photospheric and chromospheric flow fields and photospheric magnetic fields in active region NOAA 13096. The improved High-resolution Fast Imager (HiFI+) and the GREGOR Infrared Spectrograph (GRIS) acquired high-resolution imaging and spectropolarimetric data at the 1.5-meter GREGOR solar telescope at the Observatorio del Teide, Izaña, Tenerife, Spain. Background-Subtracted Activity Maps (BaSAMs) have been used to locate areas of enhanced activity, Local Correlation Tracking (LCT) provides horizontal proper motions, and near-infrared full-Stokes polarimetry offers access to magnetic fields and line-of-sight velocities. The results show that the decaying active region is characterized by a triangular region between the three leading, positive-polarity sunspots with unfavorable conditions for penumbra formation. This region has a spongy appearance in narrow-band H α images, shows signs of enhanced activity on small spatial scales, is free of divergence centers and exploding granules, lacks well-ordered horizontal flows, has low flow speeds, and is dominated by horizontal magnetic fields. Umbral cores are inactive, but the interface between pores and penumbral filaments often shows enhanced activity. Moat flows and superpenumbrae are almost always observed, when penumbral filaments are present, even in very small penumbral sectors. However, evidence of the moat flow can also be seen around pores, surviving longer than the decaying penumbral filaments. Light bridges have mainly umbral temperatures, reaching quiet-Sun temperatures in some places, show strong intensity variations, and exhibit weak photospheric horizontal flows, while narrow-band H α flow maps show substantial inflows. [ABSTRACT FROM AUTHOR]

Although many new methods have emerged in text‐driven images, the large computational power required for model training causes these methods to have a slow training process. Additionally, these methods consume a considerable amount of video random access memory (VRAM) resources during training. When generating high‐resolution images, the VRAM resources are often insufficient, which results in the inability to generate high‐resolution images. Nevertheless, recent Vision Transformers (ViTs) advancements have demonstrated their image classification and recognition capabilities. Unlike the traditional Convolutional Neural Networks based methods, ViTs have a Transformer‐based architecture, leverage attention mechanisms to capture comprehensive global information, moreover enabling enhanced global understanding of images through inherent long‐range dependencies, thus extracting more robust features and achieving comparable results with reduced computational load. The adaptability of ViTs to text‐driven image manipulation was investigated. Specifically, existing image generation methods were refined and the FastFaceCLIP method was proposed by combining the image‐text semantic alignment function of the pre‐trained CLIP model with the high‐resolution image generation function of the proposed FastFace. Additionally, the Multi‐Axis Nested Transformer module was incorporated for advanced feature extraction from the latent space, generating higher‐resolution images that are further enhanced using the Real‐ESRGAN algorithm. Eventually, extensive face manipulation‐related tests on the CelebA‐HQ dataset challenge the proposed method and other related schemes, demonstrating that FastFaceCLIP effectively generates semantically accurate, visually realistic, and clear images using fewer parameters and less time. [ABSTRACT FROM AUTHOR]

Conjugate gradient methods are widely used and attractive for large-scale unconstrained smooth optimization problems, with simple computation, low memory requirements, and interesting theoretical information on the features of curvature. Based on the strongly convergent property of the Dai–Yuan method and attractive numerical performance of the Hestenes–Stiefel method, a new hybrid descent conjugate gradient method is proposed in this paper. The proposed method satisfies the sufficient descent property independent of the accuracy of the line search strategies. Under the standard conditions, the trust region property and the global convergence are established, respectively. Numerical results of 61 problems with 9 large-scale dimensions and 46 ill-conditioned matrix problems reveal that the proposed method is more effective, robust, and reliable than the other methods. Additionally, the hybrid method also demonstrates reliable results for some image restoration problems. [ABSTRACT FROM AUTHOR]

Image restoration in adverse weather conditions is a difficult task in computer vision. In this paper, we propose a novel transformer-based framework called GridFormer which serves as a backbone for image restoration under adverse weather conditions. GridFormer is designed in a grid structure using a residual dense transformer block, and it introduces two core designs. First, it uses an enhanced attention mechanism in the transformer layer. The mechanism includes stages of the sampler and compact self-attention to improve efficiency, and a local enhancement stage to strengthen local information. Second, we introduce a residual dense transformer block (RDTB) as the final GridFormer layer. This design further improves the network's ability to learn effective features from both preceding and current local features. The GridFormer framework achieves state-of-the-art results on five diverse image restoration tasks in adverse weather conditions, including image deraining, dehazing, deraining & dehazing, desnowing, and multi-weather restoration. The source code and pre-trained models will be released. [ABSTRACT FROM AUTHOR]

In X-ray CT imaging, there are some cases where the obtained CT images have serious ring artifacts and noise, and these degraded CT images seriously affect the quality of clinical diagnosis. Thus, developing an effective method that can simultaneously suppress ring artifacts and noise is of great importance. Total variation (TV) is a famous prior regularization for image denoising in the image processing field, however, for degraded CT images, it can suppress the noise but fail to reduce the ring artifacts. To address this issue, the L 0 smoothing filter is incorporated with TV prior for CT ring artifacts and noise removal problem where the problem is transformed into several optimization sub-problems which are iteratively solved. The experiments demonstrate that the ring artifacts and noise presented in the CT image can be effectively suppressed by the proposed method and meanwhile the detailed features such as edge structure can be well preserved. As the superiority of TV and L 0 smoothing filters are fully utilized, the performance of the proposed method is better than the existing methods such as the TV-based method and L 0 -based method. [ABSTRACT FROM AUTHOR]

Image restoration is the task of aiming to obtain a high-quality image from a corrupt input image, such as deblurring and deraining. In image restoration, it is typically necessary to maintain a complex balance between spatial details and contextual information. Although a multi-stage network can optimally balance these competing goals and achieve significant performance, this also increases the system's complexity. In this paper, we propose a mountain-shaped single-stage design, which achieves the performance of multi-stage networks through a plug-and-play feature fusion middleware. Specifically, we propose a plug-and-play feature fusion middleware mechanism as an information exchange component between the encoder-decoder architectural levels. It seamlessly integrates upper-layer information into the adjacent lower layer, sequentially down to the lowest layer. Finally, all information is fused into the original image resolution manipulation level. This preserves spatial details and integrates contextual information, ensuring high-quality image restoration. Simultaneously, we propose a multi-head attention middle block as a bridge between the encoder and decoder to capture more global information and surpass the limitations of the receptive field of CNNs. In order to achieve low system complexity, we removes or replaces unnecessary nonlinear activation functions. Extensive experiments demonstrate that our approach, named as M3SNet, outperforms previous state-of-the-art models while using less than half the computational costs, for several image restoration tasks, such as image deraining and deblurring. The code and the pre-trained models will be released at https://github.com/Tombs98/M3SNet. [ABSTRACT FROM AUTHOR]

The restoration of blurred images is a crucial topic in the field of machine vision, with far-reaching implications for enhancing information acquisition quality, improving algorithmic accuracy and enriching image texture. Efforts to mitigate the phenomenon of blur have progressed from statistical approaches to those utilizing deep learning techniques. In this paper, we propose a Generative Adversarial Network (GAN)-based image restoration method to address the limitations of existing techniques in restoring color and detail in motion-blurred images. To reduce the computational complexity of generative adversarial networks and the vanishing gradient during learning, an U-net-based generator is used, and it is configured to emphasize the channel and spatial characteristics of the original information through a proposed CSAR(Channel and Spatial Attention Residual) blocks module rather than a simple concatenate operation. To validate the efficacy of the algorithm, comprehensive comparative experiments have been conducted on the GoPro dataset. Experimental results show that the peak signal-to-noise ratio is improved compared to SRN and MPRNet algorithms with good image restoration ability. Objects detection experiments using Yolo V3 showed that the proposed algorithms can generate deblerring images with higher information quality. [ABSTRACT FROM AUTHOR]

To address issues such as instability during the training of Generative Adversarial Networks, insufficient clarity in facial structure restoration, inadequate utilization of known information, and lack of attention to color information in images, a Cross-Attention Restoration Network is proposed. Initially, in the decoding part of the basic first-stage U-Net network, a combination of sub-pixel convolution and upsampling modules is employed to remedy the low-quality image restoration issue associated with single upsampling in the image recovery process. Subsequently, the restoration part of the first-stage network and the un-restored images are used to compute cross-attention in both spatial and channel dimensions, recovering the complete facial restoration image from the known repaired information. At the same time, we propose a loss function based on HSV space, assigning appropriate weights within the function to significantly improve the color aspects of the image. Compared to classical methods, this model exhibits good performance in terms of peak signal-to-noise ratio, structural similarity, and FID. [ABSTRACT FROM AUTHOR]

Image denoising is a crucial algorithm in image processing that aims to enhance image quality. Deep learning-based image denoising methods can be categorized into supervised and unsupervised approaches. Supervised learning requires pairs of noisy and noise-free training data, which is impractical in real-world scenarios. Unsupervised learning uses pairs of noisy images for training, but it may yield lower accuracy. Additionally, deep learning-based methods often require a large amount of training data. To overcome these challenges, this research proposes a self-validation Noise2Noise (SV-N2N) framework that generates validation sets using only noisy images without requiring noise-free pairs. The proposed SV-N2N method effectively reduces noise, comparable to supervised and unsupervised methods, without requiring a noise-free ground truth, which is efficient for solving real-world scenarios. [ABSTRACT FROM AUTHOR]

In this study, we introduce a novel hybrid conjugate gradient method with an adaptive strategy called asHCG method. The asHCG method exhibits the following characteristics. (i) Its search direction guarantees sufficient descent property without dependence on any line search. (ii) It possesses strong convergence for the uniformly convex function using a weak Wolfe line search, and under the same line search, it achieves global convergence for the general function. (iii) Employing the Armijo line search, it provides an approximate guarantee for worst-case complexity for the uniformly convex function. The numerical results demonstrate promising and encouraging performances in both unconstrained optimization problems and image restoration problems. [ABSTRACT FROM AUTHOR]

Image restoration is a large-scale discrete ill-posed problem, which can be transformed into a Tikhonov regularization problem that can approximate the original image. Kronecker product approximation is introduced into the Tikhonov regularization problem to produce an alternative problem of solving the generalized Sylvester matrix equation, reducing the scale of the image restoration problem. This paper considers solving this alternative problem by applying the conjugate gradient least squares (CGLS) method which has been demonstrated to be efficient and concise. The convergence of the CGLS method is analyzed, and it is demonstrated that the CGLS method converges to the least squares solution within the finite number of iteration steps. The effectiveness and superiority of the CGLS method are verified by numerical tests. [ABSTRACT FROM AUTHOR]

This study investigates the Poissonian image restoration problems. In particular, we propose a novel model that incorporates L 1 / L 2 minimization on the gradient as a regularization term combined with a box constraint and a nonlinear data fidelity term, specifically crafted to address the challenges caused by Poisson noise. We employ a splitting strategy, followed by the alternating direction method of multipliers (ADMM) to find a model solution. Furthermore, we show that under mild conditions, the sequence generated by ADMM has a sub-sequence that converges to a stationary point of the proposed model. Through numerical experiments on image deconvolution, super-resolution, and magnetic resonance imaging (MRI) reconstruction, we demonstrate superior performance made by the proposed approach over some existing gradient-based methods. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Shanghai Jiao Tong University (1006-2467) is the property of Journal of Shanghai Jiao Tong University Editorial Office and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

This paper presents a study of deep learning approaches to image recovery using spatially-extended sequential observations of near-Earth satellites. Image recovery is often a prerequisite for use of ground-based extended imagery in space domain awareness (SDA) due to aberrations induced by atmospheric turbulence along the path from satellite to sensor. Traditional deconvolution-based image recovery methods are sensitive to factors such as observation sequence length and estimates of the point spread function (PSF), which has motivated recent interest in autoencoders and other learned approaches. However, no previous study has applied general state-of-the-art image restoration models to the space domain data. In this work, we evaluate the effectiveness of recent deep learning methods, specifically Generative Adversarial Networks (GANs) and Vision Transformers, for image restoration of satellites. We analyze the trade-offs between restoration quality, time, and computational complexity of each method. We experimentally demonstrate that deep learning models provide high-quality image restoration with less data than traditional deconvolution methods. We further optimize the most successful state-of-the-art model and demonstrate its efficacy in image restoration at a previously unseen degradation level (SNIIRS = 2.5). Our deep learning models are trained on simulated data from the SILO dataset and require no training on real data, yet they restore the most severely degraded real satellite imagery with state-of-the-art performance of 27.0 dB PSNR and 0.95 SSIM on the SILO dataset, as well as better visual results on the real satellite images. [ABSTRACT FROM AUTHOR]

Recently, transformers have demonstrated notable improvements in natural advanced visual tasks. In the field of computer vision, transformer networks are beginning to supplant conventional convolutional neural networks (CNNs) due to their global receptive field and adaptability. Although transformers excel in capturing global features, they lag behind CNNs in handling fine local features, especially when dealing with underwater images containing complex and delicate structures. In order to tackle this challenge, we propose a refined transformer model by improving the feature blocks (dilated transformer block) to more accurately compute attention weights, enhancing the capture of both local and global features. Subsequently, a self-supervised method (a local and global blind-patch network) is embedded in the bottleneck layer, which can aggregate local and global information to enhance detail recovery and improve texture restoration quality. Additionally, we introduce a multi-scale convolutional block attention module (MSCBAM) to connect encoder and decoder features; this module enhances the feature representation of color channels, aiding in the restoration of color information in images. We plan to deploy this deep learning model onto the sensors of underwater robots for real-world underwater image-processing and ocean exploration tasks. Our model is named the refined transformer combined with convolutional block attention module (RT-CBAM). This study compares two traditional methods and six deep learning methods, and our approach achieved the best results in terms of detail processing and color restoration. [ABSTRACT FROM AUTHOR]

The development of the Internet of Things has led to a surge in edge devices. The image detection algorithm, one of the commonly used algorithms in edge computing, is affected by environments such as weather, light, air humidity, smoke, and dust, so an image restoration algorithm is needed to preprocess the image in practice. Most currently proposed deep learning image restoration algorithms are based on general-purpose servers with high computational overhead to minimize the environmental effects. Edge devices are limited in size, power consumption, and computing performance, making the performance of deep learning-based image restoration algorithms on edge devices poor. In this work, we propose an image restoration algorithm that combines wavelet transform and transformer, named MWformer, to reduce computational overhead, optimize the feature map size, network structure, and network depth, and introduce the wavelet transformation to reduce the super-parameters. Experimental tests on multiple public datasets for various image restoration tasks show that the proposed MWformer ensures high performance in numerous image restoration tasks, and the computational overhead is 10% of the state-of-the-art algorithm, on average. [ABSTRACT FROM AUTHOR]

RainDNet is an advanced image deraining model that refines the "Multi-Stage Progressive Image Restoration Network" (MPRNet) for superior computational efficiency and perceptual fidelity. RainDNet's innovative architecture employs depthwise separable convolutions instead of MPRNet's traditional ones, reducing model complexity and improving computational efficiency while preserving the feature extraction ability. RainDNet's performance is enhanced by a multi-objective loss function combining perceptual loss for visual quality and Structural Similarity Index Measure (SSIM) loss for structural integrity. Experimental evaluations demonstrate RainDNet's superior performance over MPRNet in terms of Peak Signal-to-Noise Ratio (PSNR), SSIM, and BRISQUE (Blind Referenceless Image Spatial Quality Evaluator) scores across multiple benchmark datasets, underscoring its aptitude for maintaining image fidelity while restoring structural and textural details. Our findings invite further explorations into more efficient architectures for image restoration tasks, contributing significantly to the field of computer vision. Ultimately, RainDNet lays the foundation for future, resource-efficient image restoration models capable of superior performance under diverse real-world scenarios. [ABSTRACT FROM AUTHOR]

Although existing object detectors achieve encouraging performance of object detection and localisation under real ideal conditions, the detection performance in adverse weather conditions (snowy) is very poor and not enough to cope with the detection task in adverse weather conditions. Existing methods do not deal well with the effect of snow on the identity of object features or usually ignore or even discard potential information that can help improve the detection performance. To this end, the authors propose a novel and improved end‐to‐end object detection network joint image restoration. Specifically, in order to address the problem of identity degradation of object detection due to snow, an ingenious restoration‐detection dual branch network structure combined with a Multi‐Integrated Attention module is proposed, which can well mitigate the effect of snow on the identity of object features, thus improving the detection performance of the detector. In order to make more effective use of the features that are beneficial to the detection task, a Self‐Adaptive Feature Fusion module is introduced, which can help the network better learn the potential features that are beneficial to the detection and eliminate the effect of heavy or large local snow in the object area on detection by a special feature fusion, thus improving the network's detection capability in snowy. In addition, the authors construct a large‐scale, multi‐size snowy dataset called Synthetic and Real Snowy Dataset (SRSD), and it is a good and necessary complement and improvement to the existing snowy‐related tasks. Extensive experiments on a public snowy dataset (Snowy‐weather Datasets) and SRSD indicate that our method outperforms the existing state‐of‐the‐art object detectors. [ABSTRACT FROM AUTHOR]

The Douglas-Rachford method (DR) is one of the most computationally efficient iterative methods for the large scale linear systems of equations. Based on the randomized alternating reflection and relaxation strategy, we propose a randomized block Douglas–Rachford method for solving the matrix equation A X B = C . The Polyak's and Nesterov-type momentums are integrated into the randomized block Douglas–Rachford method to improve the convergence behaviour. The linear convergence of the resulting algorithms are proven. Numerical simulations and experiments of randomly generated data, real-world sparse data, image restoration problem and tensor product surface fitting in computer-aided geometry design are performed to illustrate the feasibility and efficiency of the proposed methods. [ABSTRACT FROM AUTHOR]

Variational regularization, renowned for its sound theoretical foundations and impressive performance, is widely used in image restoration. The traditional regularization models typically use a predefined regularizer to promote smoothness in the solution. However, these models do not explicitly take into account any external information that should be preserved in the restoration. In this paper, we introduce a novel guided regularization model to enhance the efficacy of traditional regularization. Our model incorporates an external guidance regularizer, utilizing a guidance image to bolster the quality of restoration. By integrating this external information into the regularization process, the model is better equipped to preserve specific features or attributes indicated by the guidance image, leading to more accurate and aesthetically pleasing restored images. Furthermore, we demonstrate the convexity of the model and prove the existence and uniqueness of the solution. The alternating direction method of multipliers (ADMM) algorithm is employed to numerically solve the proposed model. In the experimental evaluation, the proposed model is applied to image denoising and deblurring tasks. The experiments successfully validate the proposed model and algorithm. Compared with several state-of-the-art models, the proposed model demonstrates the best performance in terms of peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM). • A novel guided regularization model is proposed, which aims to incorporate additional guidance into regularization process. • We demonstrate the convexity of the model and prove the existence and uniqueness of the solution. • The proposed guided regularization model is applied to image denoising and deblurring tasks, yielding satisfactory results. [ABSTRACT FROM AUTHOR]

When photographing through a piece of glass, reflections usually degrade the quality of captured images or videos. In this paper, by exploiting periodically varying light flickering, we investigate the problem of removing strong reflections from contaminated image sequences or videos with a unified capturing setup. We propose a learning-based method that utilizes short-term and long-term observations of mixture videos to exploit one-side contextual clues in fluctuant components and brightness-consistent clues in consistent components for achieving layer separation and flickering removal, respectively. A dataset containing synthetic and real mixture videos with light flickering is built for network training and testing. The effectiveness of the proposed method is demonstrated by the comprehensive evaluation on synthetic and real data, the application for video flickering removal, and the exploratory experiment on high-speed scenes. [ABSTRACT FROM AUTHOR]

Under low‐light imaging conditions, visual scenes captured by intelligent waterborne transportation systems often suffer from low‐intensity illumination and noise corruption. The visual quality degradation would lead to negative effects in maritime surveillance, e.g., vessel detection, positioning and tracking, etc. To restore the low‐light images, we develop an effective visibility enhancement method, which contains a coarse‐to‐fine framework of spatially‐smooth illumination estimation. In particular, the refined illumination is effectively generated by optimizing a novel structure‐preserving variational model on the coarse version, estimated through the Max‐RGB method. The proposed variational model has the capacity of suppressing the textural details while preserving the main structures in the refined illumination map. To further boost imaging performance, the refined illumination is adjusted through the Gamma correction to increase brightness in dark regions. We then estimate the refined reflection map by implementing the joint denoising and detail boosting strategies on the original reflection. In this work, the original reflection is yielded by dividing the input image using the refined illumination. We finally produce the enhanced image by multiplying the adjusted illumination and the refined reflection. Experiments on synthetic and realistic datasets illustrate that our method can achieve comparable results to the state‐of‐the‐art techniques under different imaging conditions. [ABSTRACT FROM AUTHOR]

This paper introduces an improved fuzzy association memory (IFAM), an advanced FAM method based on the T-conorm probability operator. Specifically, the T-conorm probability operator fuzzifies the input data and performs fuzzy logic operations, effectively handling ambiguity and uncertainty during image restoration, which enhances the accuracy and effectiveness of the restoration results. Experimental results validate the performance of IFAM by comparing it with existing fuzzy association memory techniques. The root mean square error shows that the restoration rate of IFAM reached 80%, compared to only 40% for the traditional fuzzy association memory technique. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Shenzhen University Science & Engineering is the property of Editorial Department of Journal of Shenzhen University Science & Engineering and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The red revolutionary heritage is a valuable part of China's historical and cultural legacy, with the potential to generate economic benefits through its thoughtful development. However, challenges such as insufficient understanding, lack of comprehensive planning and layout, and limited protection and utilization methods hinder the full realization of the political, cultural, and economic value of red heritage. To address these problems, this paper thoroughly examines the current state of red revolutionary heritage protection and identifies the problems within the preservation process. Moreover, it proposes leveraging advanced artificial intelligence (AI) technology to repair some damaged image data. Specifically, this paper introduces a red revolutionary cultural relic image-restoration model based on a generative adversarial network (GAN). This model was trained using samples of damaged image and utilizes high-quality models to restore these images effectively. The study also integrates real-world revolutionary heritage images for practical application and assesses its effectiveness through questionnaire surveys. The survey results show that AI algorithms and image-processing technologies hold significant potential in the protection of revolutionary heritage. [ABSTRACT FROM AUTHOR]

Abstract In response to the inadequate utilization of prior information in current mural inpainting processes, leading to issues such as semantically unreliable inpaintings and the presence of artifacts in the inpainting area, a Dunhuang mural inpainting method based on reference guidance and multi‐scale feature fusion is proposed. First, the simulated broken mural, the mask image, and the reference mural are input into the model to complete the multi‐level embedding of patches and align the multi‐scale fine‐grained features of damaged murals and reference murals. Following the patch embedding module, a hybrid residual module is added based on hybrid attention to fully extract mural features. In addition, by continuing the residual concatenation of outputs of the hierarchical embedding module improves the ability of the model to represent deeper features, and improves the robustness and generalisation of the model. Second, the encoded features are fed into the decoder to generate decoded features. Finally, the convolutional tail is employed to propagate them and complete the mural painting. Experimental validation on the Dunhuang mural dataset demonstrates that, compared to other algorithms, this model exhibits higher evaluation metrics in the inpainting of extensively damaged murals and demonstrates overall robustness. In terms of visual effects, the results of this model in the inpainting process exhibit finer textures, richer semantic information, more coherent edge structures, and a closer resemblance to authentic murals.

Image restoration under adverse weather conditions is of great significance for the subsequent advanced computer vision tasks. However, most existing image restoration algorithms only remove single weather degradation, and few studies has been conducted on all-in-one weather-degraded image restoration. The denoising diffusion probability model is combined with Vision Transformer to propose a Transformer-based diffusion model for all-in-one weather-degraded image restoration. First, the weather-degraded image is utilized as the condition to guide the reverse sampling of diffusion model and generate corresponding clean background image. Then, the subspace transposed Transformer for noise estimation (NE-STT) is proposed, which utilizes the degraded image and the noisy state to estimate noise distribution, including the subspace transposed self-attention (STSA) mechanism and a dual grouped gated feed-forward network (DGGFFN). The STSA adopts subspace transformation coefficient to effectively capture global long-range dependencies while significantly reducing computational burden. The DGGFFN employs the dual grouped gated mechanism to enhance the nonlinear characterization ability of feed-forward network. The experimental results show that in comparison with the recently developed algorithms, such as All-in-One and TransWeather, the method proposed obtains a performance gain of 3.68 and 3.08 dB in average peak signal-to-noise ratio while 2.93% and 3.13% in average structural similarity on 5 weather-degraded datasets.

Abstract Although many new methods have emerged in text‐driven images, the large computational power required for model training causes these methods to have a slow training process. Additionally, these methods consume a considerable amount of video random access memory (VRAM) resources during training. When generating high‐resolution images, the VRAM resources are often insufficient, which results in the inability to generate high‐resolution images. Nevertheless, recent Vision Transformers (ViTs) advancements have demonstrated their image classification and recognition capabilities. Unlike the traditional Convolutional Neural Networks based methods, ViTs have a Transformer‐based architecture, leverage attention mechanisms to capture comprehensive global information, moreover enabling enhanced global understanding of images through inherent long‐range dependencies, thus extracting more robust features and achieving comparable results with reduced computational load. The adaptability of ViTs to text‐driven image manipulation was investigated. Specifically, existing image generation methods were refined and the FastFaceCLIP method was proposed by combining the image‐text semantic alignment function of the pre‐trained CLIP model with the high‐resolution image generation function of the proposed FastFace. Additionally, the Multi‐Axis Nested Transformer module was incorporated for advanced feature extraction from the latent space, generating higher‐resolution images that are further enhanced using the Real‐ESRGAN algorithm. Eventually, extensive face manipulation‐related tests on the CelebA‐HQ dataset challenge the proposed method and other related schemes, demonstrating that FastFaceCLIP effectively generates semantically accurate, visually realistic, and clear images using fewer parameters and less time.

Abstract The tensor singular value decomposition‐based model has garnered increasing attention in addressing tensor recovery challenges. However, existing tensor recovery methods exhibit certain inherent limitations. Some ignore the simultaneous effects of noise and missing values, while most can't handle higher‐order tensors, which are not reflective of real‐world scenarios. The information redundancy within tensor data often leads to a prevailing low‐rank structure, making low‐rankness a vital prior in the tensor recovery process. To tackle this pressing issue, a robust low‐rank tensor recovery framework is proposed to rehabilitate higher‐order tensors corrupted by sparse noise and missing entries. In the model, the tensor nuclear norm derived for order‐d tensors (d ≥ 4) are employed as a representation of the low‐rank prior, while utilizing the L1‐norm to model the sparse noise. To solve the proposed model, an efficient Alternating direction method of multipliers algorithm is developed. A series of experiments are performed on synthetic and real‐world datasets. The results show that the superior performance of the method compared with other algorithms dedicated to addressing order‐d tensor recovery challenges. Notably, in scenarios where the data is severely compromised (noise ratio 40%, sample ratio 70%), the algorithm consistently outperforms its competitors, achieving significantly improved results.

Abstract Image super‐resolution (SR) has been widely applied in remote sensing to generate high‐resolution (HR) images without increasing hardware costs. However, SR is a severe ill‐posed problem. As deep learning advances, existing methods have solved this problem to a certain extent. However, the complex spatial distribution of remote sensing images still poses a challenge in effectively extracting abundant high‐frequency details from the images. Here, a single‐image super‐resolution (SISR) network based on the generative adversarial network (GAN) for remote sensing is presented, called JOA‐GAN. Firstly, a joint‐attention module (JOA) is proposed to focus the network on high‐frequency regions in remote sensing images to enhance the quality of image reconstruction. In the generator network, a multi‐scale densely connected feature extraction block (ERRDB) is proposed, which acquires features at different scales using MSconv blocks containing multi‐scale convolutions and automatically adjusts the features by JOA. In the discriminator network, the relative discriminator is used to compute the relative probability instead of the absolute probability, which helps the network learn clearer and more realistic texture details. JOA‐GAN is compared with other advanced methods, and the results demonstrate that JOA‐GAN has improved objective evaluation metrics and achieved superior visual effects.

Abstract Face shape priors such as landmarks, heatmaps, and parsing maps are widely used to improve face super resolution (SR). It is observed that face priors provide locations of high‐frequency details in key facial areas such as the eyes and mouth. However, existing methods fail to effectively exploit the high‐frequency information by using the priors as either constraints or inputs. This paper proposes a novel high frequency highway (H2F) framework to better utilize prior information for face SR, which dynamically decomposes the final SR face into a coarse SR face and a high frequency (HF) face. The coarse SR face is reconstructed from a low‐resolution face via a texture branch, using only pixel‐wise reconstruction loss. Meanwhile, the HF face is directly generated from face priors via an HF branch that employs the proposed inception–hourglass model. As a result, H2F allows the face priors to have a direct impact on the SR face by adding the outputs of both branches as the final result and provides an extra face editing function. Extensive experiments show that H2F significantly outperforms state‐of‐the‐art face SR methods, is general for different texture branch models and face priors, and is robust to dataset mismatch and pose variations.

Abstract Although existing object detectors achieve encouraging performance of object detection and localisation under real ideal conditions, the detection performance in adverse weather conditions (snowy) is very poor and not enough to cope with the detection task in adverse weather conditions. Existing methods do not deal well with the effect of snow on the identity of object features or usually ignore or even discard potential information that can help improve the detection performance. To this end, the authors propose a novel and improved end‐to‐end object detection network joint image restoration. Specifically, in order to address the problem of identity degradation of object detection due to snow, an ingenious restoration‐detection dual branch network structure combined with a Multi‐Integrated Attention module is proposed, which can well mitigate the effect of snow on the identity of object features, thus improving the detection performance of the detector. In order to make more effective use of the features that are beneficial to the detection task, a Self‐Adaptive Feature Fusion module is introduced, which can help the network better learn the potential features that are beneficial to the detection and eliminate the effect of heavy or large local snow in the object area on detection by a special feature fusion, thus improving the network's detection capability in snowy. In addition, the authors construct a large‐scale, multi‐size snowy dataset called Synthetic and Real Snowy Dataset (SRSD), and it is a good and necessary complement and improvement to the existing snowy‐related tasks. Extensive experiments on a public snowy dataset (Snowy‐weather Datasets) and SRSD indicate that our method outperforms the existing state‐of‐the‐art object detectors.

Abstract Under low‐light imaging conditions, visual scenes captured by intelligent waterborne transportation systems often suffer from low‐intensity illumination and noise corruption. The visual quality degradation would lead to negative effects in maritime surveillance, e.g., vessel detection, positioning and tracking, etc. To restore the low‐light images, we develop an effective visibility enhancement method, which contains a coarse‐to‐fine framework of spatially‐smooth illumination estimation. In particular, the refined illumination is effectively generated by optimizing a novel structure‐preserving variational model on the coarse version, estimated through the Max‐RGB method. The proposed variational model has the capacity of suppressing the textural details while preserving the main structures in the refined illumination map. To further boost imaging performance, the refined illumination is adjusted through the Gamma correction to increase brightness in dark regions. We then estimate the refined reflection map by implementing the joint denoising and detail boosting strategies on the original reflection. In this work, the original reflection is yielded by dividing the input image using the refined illumination. We finally produce the enhanced image by multiplying the adjusted illumination and the refined reflection. Experiments on synthetic and realistic datasets illustrate that our method can achieve comparable results to the state‐of‐the‐art techniques under different imaging conditions.

In the field of traditional landscape painting and art image restoration, traditional restoration methods have gradually revealed limitations with the development of society and technological progress. In order to enhance the restoration effects of Chinese landscape paintings, an innovative image restoration algorithm is designed in this research, combining edge restoration with generative adversarial networks (GANs). Simultaneously, a novel image restoration model with embedded multi-scale attention dilated convolution is proposed to enhance the modeling capability for details and textures in landscape paintings. To better preserve the structural features of artistic images, a structural information-guided art image restoration model is introduced. The introduction of adversarial networks into the repair model can improve the repair effect. The art image repair model adds a multi-scale attention mechanism to handle more complex works of art. The research results show that the image detection model improves by 0.20, 0.07, and 0.06 in the Spearman rank correlation coefficient, Pearson correlation coefficient, and peak signal-to-noise ratio (PSNR), respectively, compared to other models. The proposed method outperforms mean filtering, wavelet denoising, and median filtering algorithms by 6.3, 9.1, and 15.8 dB in PSNR and by 0.06, 0.12, and 0.11 in structural similarity index. In the image restoration task, the structural similarity and information entropy indicators of the research model increase by approximately 9.3 and 3%, respectively. The image restoration method proposed in this study is beneficial for preserving and restoring precious cultural heritage, especially traditional Chinese landscape paintings, providing new technological means for cultural relic restoration.

In this paper, we explore an algorithm for low-light image enhancement based on Retinex modules. A low light enhancement network based on Ghost-Block and unique image decomposition is proposed, termed as GDNet. This addresses the problem of color distortion that exists in current enhancement algorithms. Firstly, a special module called Ghost-Block is proposed which can effectively reduce the redundant features in the network. Secondly, we design a unique decomposition network based on the Ghost-Block. It decomposes low-light image into texture map, color map, and illumination map, representing the texture structure, color, and illumination distribution in the original image, respectively. Taking into account the influence of different illumination scenarios on color information in the subsequent adjustment process, we develop a coupled color and illumination adjustment network. This treats color and illumination adjustment in low-light enhancement as a joint optimization problem rather than separate sub-tasks, aiming to achieve a natural color distribution with balanced illumination in the enhanced image. Finally, considering potential loss of texture detail information during the light adjustment process, we design a texture adjustment network to further restore the texture structure in natural lighting scenes. Extensive experiments demonstrate that our algorithm outperforms state-of-the-art methods in terms of light adjustment and color fidelity. Moreover, our proposed algorithm exhibits a high degree of similarity with Ground Truth in experiments using normal lighting images as a test set. The source code will be released at . [ABSTRACT FROM AUTHOR]

We present a novel approach to leverage prior knowledge encapsulated in pre-trained text-to-image diffusion models for blind super-resolution. Specifically, by employing our time-aware encoder, we can achieve promising restoration results without altering the pre-trained synthesis model, thereby preserving the generative prior and minimizing training cost. To remedy the loss of fidelity caused by the inherent stochasticity of diffusion models, we employ a controllable feature wrapping module that allows users to balance quality and fidelity by simply adjusting a scalar value during the inference process. Moreover, we develop a progressive aggregation sampling strategy to overcome the fixed-size constraints of pre-trained diffusion models, enabling adaptation to resolutions of any size. A comprehensive evaluation of our method using both synthetic and real-world benchmarks demonstrates its superiority over current state-of-the-art approaches. Code and models are available at https://github.com/IceClear/StableSR. [ABSTRACT FROM AUTHOR]

Haze is a natural phenomenon that negatively affects image clarity and quality, posing challenges across various image-related applications. Traditional dehazing models often suffer from overfitting when trained on synthetic hazy-clean image pairs, which do not generalize well to real-world hazy conditions. To tackle this, recent methodologies have explored training models on unpaired data, better reflecting the variability encountered in natural scenes. This dual capability of CycleGAN is particularly beneficial for overcoming the overfitting issues associated with synthetic datasets. By incorporating CycleGAN into our DehazeDNet framework, we ensure that our dehazing model not only translates images effectively but also respects the physical characteristics of haze. Inspired by the D4 model, our approach includes a Depth Evaluation Block to estimate scene depth from images. Since haze density often correlates with scene depth, this depth information is crucial for accurate haze modeling. We utilize the U-Net architecture for the Depth Evaluation Block due to its proven efficiency in image-to-image translation tasks. To preserve the accuracy of the dehazed images, we incorporate an identity loss function into our model. Identity loss ensures that the dehazed output retains the essential characteristics of the input image. Our results demonstrate an increase in SSIM and PSNR compared to other unsupervised dehazing models, highlighting the efficiency of our method in maintaining image quality and details while removing haze. [ABSTRACT FROM AUTHOR]