SRGAT: Single Image Super-Resolution With Graph Attention Network

Yan, Yanyang; Ren, Wenqi; Hu, Xiaobin; Li, Kun; Shen, Haifeng; Cao, Xiaochun

doi:10.1109/tip.2021.3077135

Cited by 38 publications

(7 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition, it was concluded that the diagnostic sensitivity (95.34%), specificity (75%), and accuracy (94.44%) of SR-CNN algorithm-based MRI images were obviously superior to those of conventional MRI (81.40%, 50%, and 80%). The conclusion revealed that SR-CNN algorithm-based MRI images could improve the diagnostic effect of MRI images, which was consistent with the outcomes of the studies conducted by Yan et al [25] and Park et al [26] and provided the basis for the accuracy of subsequent studies. Based on the above research results, the effect of TC on children with SVE was investigated and compared with conventional nursing effect.…”

Section: Discussionsupporting

confidence: 85%

Influences of Magnetic Resonance Imaging Superresolution Algorithm-Based Transition Care on Prognosis of Children with Severe Viral Encephalitis

Wang

Zhang

2022

Computational and Mathematical Methods in Medicine

View full text Add to dashboard Cite

Objective. Its goal was to see how convolutional neural network- (CNN-) based superresolution (SR) technology magnetic resonance imaging- (MRI-) assisted transition care (TC) affected the prognosis of children with severe viral encephalitis (SVE) and how effective it was. Methods. 90 SVE children were selected as the research objects and divided into control group (39 cases receiving conventional nursing intervention) and observation group (51 cases performed with conventional nursing intervention and TC intervention) according to their nursing purpose. Based on SR-CNN-optimized MRI images, diagnosis was implemented. Life treatment and sequelae in two groups were compared. Results. After the processing by CNN algorithm-based SR, peak signal to noise ratio (PSNR) (40.08 dB) and structural similarity (SSIM) (0.98) of MRI images were both higher than those of fully connected neural network (FNN) (38.01 dB, 0.93) and recurrent neural network (RNN) (37.21 dB, 0.93) algorithms. Diagnostic sensitivity (95.34%), specificity (75%), and accuracy (94.44%) of MRI images were obviously superior to those of conventional MRI (81.40%, 50%, and 80%). PedsQLTM 4.0 scores of the observation group 1 to 3 months after discharge were all higher than those of the control group ( 54.55 ± 5.76 vs. 52.32 ± 5.12 and 66.32 ± 8.89 vs. 55.02 ± 5.87 ). Sequela incidence in the observation group (13.73%) was apparently lower than that in the control group (43.59%) ( P < 0.05 ). Conclusion. (1) SR-CNN algorithm could increase the definition and diagnostic ability of MRI images. (2) TC could reduce sequelae incidence among SVE children and improve their quality of life (QOL).

show abstract

Section: Discussionsupporting

confidence: 85%

Influences of Magnetic Resonance Imaging Superresolution Algorithm-Based Transition Care on Prognosis of Children with Severe Viral Encephalitis

Wang

Zhang

2022

Computational and Mathematical Methods in Medicine

View full text Add to dashboard Cite

show abstract

“…Step 3: For each target patch x i in X 0 , utilize steer kernel regression to estimate matrix T i , perform shape-adaptive grouping on X 0 and calculate each low-rank matrix T i 􏽥 X i via (15). Step 4: Obtain the final reconstructed HR image 􏽢 X� X H via (16).…”

Section: Experiments On Noisymentioning

confidence: 99%

“…where D refers to a downsampling operator, H represents a blur matrix, X and Y, respectively, denote the original HR image and the degraded LR image, and v is additive Gaussian noise. Generally, existing SISR methods can be divided into interpolation-based approaches [1][2][3], reconstruction-based approaches [4][5][6][7], and learning-based approaches [8][9][10][11][12][13][14][15][16][17].…”

Section: Introductionmentioning

confidence: 99%

“…eir crucial idea is the discovery of the mapping relationship between LR and HR exemplar image patch pairs. According to the establishment of a mapping relationship, learning-based methods generally include neighbor-based approaches [14,15,18,19], sparse representation-based approaches [8,[20][21][22][23][24][25], example regression-based approaches [9,10,12,13], and deep learning-based approaches [11,16,17,[26][27][28][29][30][31][32].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Locally Regularized Collaborative Representation and an Adaptive Low-Rank Constraint for Single Image Superresolution

Gao

Cheng

Chen

et al. 2022

Mobile Information Systems

View full text Add to dashboard Cite

Learning-based superresolution reconstruction is an efficient image processing technique that has become a popular topic in recent years. Since superresolution is an ill-conditioned problem, appropriate image priors or examples are the key factors for recovering high-quality images with rich details. However, in some current advanced superresolution methods, the mappings established using dictionaries of low- and high-resolution examples cannot effectively reflect the relationship between the low- and high-resolution spaces. Therefore, we introduce a local structure prior to the collaborative representation to constrain the projection matrix; this structure can better represent the nonlinear mapping between the low- and the high-resolution feature spaces. Then, based on the redundancy of similar image patches, a shape-adaptive low-rank constraint is utilized to explore the images’ nonlocal self-similarity, and the local and nonlocal priors complement each other to enhance the recovered image quality. Finally, an iterative optimization algorithm is adopted to solve our proposed superresolution model. Numerous experiments were performed to verify the proposed method, and the results demonstrate its superiority to some state-of-the-art methods both quantitatively and qualitatively.

show abstract

“…However, most CNN-based SR methods usually treat different feature contents equally and lack the ability discriminate the desired high-frequency details. In recent years, some researchers proposed to apply attention mechanism to enhance the discriminative representation of the network and achieved good experimental performance, which has been widely used in many computer vision tasks [25][26][27]. Hu et al [28] proposed the squeeze-and-excitation block to learn the channel-wise information and improve the representation ability of the model.…”

Section: Introductionmentioning

confidence: 99%

Pyramidal dense attention networks for single image super‐resolution

Gui

Zhang

et al. 2022

IET Image Processing

View full text Add to dashboard Cite

Recently, residual and dense networks have effectively promoted the development of image super-resolution (SR). However, most dense networks based SR methods do not make full use of dense feature information. To solve this problem, a pyramidal dense attention network for single image super-resolution is proposed in this paper. In this method, the proposed pyramidal dense learning can gradually increase the width of the densely connected layer inside a pyramidal dense block to extract deep features efficiently. Meanwhile, the adaptive group convolution that the number of groups grows linearly with dense convolutional layers is introduced to relieve the parameter explosion. Besides, a novel joint attention to capture cross-dimension interaction between the spatial dimensions and channel dimension in an efficient way for providing rich discriminative feature representations is also proposed. Extensive experimental results show that the method achieves comparable performance in comparison with the state-of-the-art SR methods.

show abstract

SRGAT: Single Image Super-Resolution With Graph Attention Network

Cited by 38 publications

References 43 publications

Influences of Magnetic Resonance Imaging Superresolution Algorithm-Based Transition Care on Prognosis of Children with Severe Viral Encephalitis

Influences of Magnetic Resonance Imaging Superresolution Algorithm-Based Transition Care on Prognosis of Children with Severe Viral Encephalitis

Locally Regularized Collaborative Representation and an Adaptive Low-Rank Constraint for Single Image Superresolution

Pyramidal dense attention networks for single image super‐resolution

Contact Info

Product

Resources

About