Image super-resolution: A comprehensive review, recent trends, challenges and applications

Lepcha, Dawa Chyophel; Goyal, Bhawna; Dogra, Ayush; Goyal, Vishal

doi:10.1016/j.inffus.2022.10.007

Cited by 105 publications

(28 citation statements)

References 92 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The experimental results show that our method achieves superior performance compared with other models in terms of both PSNR and SSIM. In addition, the following points can be summarized: (1) In terms of PSNR and SSIM, except in a few rare cases, DBFRN+ and DBFRN have optimal and suboptimal performance on benchmark datasets, respectively. (2) Although the overall number of parameters…”

Section: Benchmark Resultsmentioning

confidence: 99%

“…By reviewing past studies [9,10], we can easily observe that there is a strong correlation among different scales in SR, which is mainly manifested in two aspects: (1) Compared to multiple single-scale models, multi-scale SR models are more compact because they share most of the parameters across different scales while maintaining accuracy. For example, the performance of a multi-scale deep super-resolution network (MDSR) [9] is comparable to that of an enhanced deep super-resolution network (EDSR) [9], but the former has fewer parameters on all scales.…”

Section: Introductionmentioning

confidence: 99%

“…Single image super‐resolution (SR) [1] is a fundamental vision task, which aims to reconstruct high‐resolution (HR) images at different scales from a given low‐resolution (LR) image. The research in this domain holds the potential for diverse applications, including surveillance imaging [2], medical imaging [3], and remote sensing imaging [4], etc.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Diverse branch feature refinement network for efficient multi‐scale super‐resolution

Zhang,

Lei

et al. 2024

IET Image Processing

View full text Add to dashboard Cite

Despite the existence of various super‐resolution (SR) methods, most of them focus on designing models for specific upscaling factors rather than fully exploiting inter‐scale correlation to improve efficiency. In contrast, multi‐scale SR methods can effectively reduce the redundancy of network parameters by aggregating the feature extraction processes corresponding to multiple scales into a unified process. The aim of this study is to enhance the compactness and efficiency of the SR model. Thus, an efficient multi‐scale SR method called the diverse branch feature refinement network (DBFRN) is proposed. By decoupling the training process and inference process based on the idea of structural re‐parameterization, multi‐branch topology is adopted to enrich multi‐scale learning and merge branches to achieve efficient inference with equivalent effects. Specifically, two re‐parameterization strategies are designed and two corresponding feature refinement blocks for different feature levels in multi‐scale SR network. Extensive experiments demonstrate that the proposed multi‐scale SR method is effective and efficient, and it can outperform advanced single‐scale methods in terms of quantity and quality.

show abstract

Section: Benchmark Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Diverse branch feature refinement network for efficient multi‐scale super‐resolution

Zhang,

Lei

et al. 2024

IET Image Processing

View full text Add to dashboard Cite

show abstract

“…Single image super-resolution (SR) 1 is a long-standing challenging task whose goal is to reconstruct a high-resolution (HR) image from its low-resolution (LR) counterpart. Due to the uncertainty and irreversibility of the image degradation process, it is a typical ill-posed problem in computer vision.…”

Section: Introductionmentioning

confidence: 99%

Global context-aware feature modulation networks for unified multi-scale super-resolution

et al. 2023

View full text Add to dashboard Cite

.Compared with training multiple single-scale models to achieve different scale image super-resolution (SR), using the internal correlation of multiple scales to aggregate multi-scale learning into a single model can effectively reduce the network parameters and improve efficiency. Correspondingly, it is important to study how to take the advantage of inter-scale correlation at a fixed model size for the performance improvement of multi-scale SR. We combine the capture of richer context dependencies and convolutional response modulation to improve the discriminative ability of the model, and then propose a global context-aware feature modulation network, which has a stronger representational capability for multi-scale SR. Specifically, we first model the global context based on the self-attention mechanism, which can effectively extract rich global information from convolutional features. Then, we propose to modulate the feature responses using global contextual dependencies in spatial and channel dimensions respectively. Further, we construct the global context-aware feature modulation block, which can be stacked to form a deep feature modulation network and applied to the unified multi-scale SR architecture to learn inter-scale correlation. Extensive experiments demonstrate that the proposed global context-aware modulation is effective, which can improve the performance of multi-scale SR task both quantitatively and qualitatively.

show abstract

“…Deep learning techniques continue to outperform traditional based algorithms in terms of efficiency and effectiveness due to end to end training (Bashir et al, 2021). Deep learning algorithms carry flexibility which could handle super resolution issue with different scale factors, blur kernels and noise levels inside a unified maximum a posteriori framework (Dawa et al, 2023). Overall, the ESPCN and ESRGAN models are currently the most advanced and widely-used SR models.…”

Section: Introductionmentioning

confidence: 99%

Super-resolution reconstruction of sea surface pollutant diffusion images based on deep learning models: a case study of thermal discharge from a coastal power plant

Duan

Liu

2023

Front. Mar. Sci.

View full text Add to dashboard Cite

While remote sensing images could convey essential information of surface water environment, the low spatial resolution limits their application. This study carried out a series of experiment tests of thermal discharge from a coastal power plant and constructed an image dataset HY_IRS, representing the transport and diffusion of discharged heated water in tidal waters. Two image super-resolution (SR) reconstruction models based on deep learning (DL), ESPCN and ESRGAN, were trained based on this dataset and then used to reconstruct high-resolution remote sensing images. It shows that the two DL models can markedly improve the spatial resolution of the surface diffusion image of thermal discharging, with the PSNR improved by 8.3% on average. The trained two models were successfully used to improve the spatial resolution of thermal infrared remote sensing SST images from Landsat8 TIRS, indicating that the SR model based on DL has a good effect and a crucial application prospect in the field of improving the resolution of pollutant diffusion remote sensing images.

show abstract

Image super-resolution: A comprehensive review, recent trends, challenges and applications

Cited by 105 publications

References 92 publications

Diverse branch feature refinement network for efficient multi‐scale super‐resolution

Diverse branch feature refinement network for efficient multi‐scale super‐resolution

Global context-aware feature modulation networks for unified multi-scale super-resolution

Super-resolution reconstruction of sea surface pollutant diffusion images based on deep learning models: a case study of thermal discharge from a coastal power plant

Contact Info

Product

Resources

About