Learning a Single Convolutional Super-Resolution Network for Multiple Degradations

By Kai Zhang, Wangmeng Zuo, Lei Zhang. 2017

1,059 citations recorded by the source corpus.

Recent years have witnessed the unprecedented success of deep convolutional neural networks (CNNs) in single image super-resolution (SISR). However, existing CNN-based SISR methods mostly assume that a low-resolution (LR) image is bicubicly downsampled from a high-resolution (HR) image, thus inevitably giving rise to poor performance when the true degradation does not follow this assumption. Moreover, they lack scalability in learning a single model to nonblindly deal with multiple degradations. To address these issues, we propose a general framework with dimensionality stretching strategy that enables a single convolutional super-resolution network to take two key factors of the SISR degradation process, i.e., blur kernel and noise level, as input. Consequently, the super-resolver can handle multiple and even spatially variant degradations, which significantly improves the practicability. Extensive experimental results on synthetic and real LR images show that the proposed convolutional super-resolution network not only can produce favorable results on multiple degradations but also is computationally efficient, providing a highly effective and scalable solution to practical SISR applications.

Subfields: Computer Vision and Pattern Recognition, Media Technology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

Designing a Practical Degradation Model for Deep Blind Image Super-Resolution (2021, methodology): “[3] propose to estimate the blur kernel via an internal-GAN method before applying the non-blind ZSSR [45] and SRMD [58] methods.”
Image Super-Resolution Using Very Deep Residual Channel Attention Networks (2018, methodology): “LapSRN [6] MSLapSRN [7] ENet-PAT [8] MemNet [9] EDSR [10] SRMDNF [11] RCAN (ours)”
Toward Real-World Single Image Super-Resolution: A New Benchmark and a New Model (2019, methodology): “SISR results generated by (b) bicubic interpolator, RCAN models [64] trained using image pairs (in DIV2K [46]) with (c) bicubic degradation (BD), (d) multiple simulated degradations (MD) [62], and (e) authentic distortions in our RealSR dataset.”
Correction to: Single Image Super-Resolution via a Holistic Attention Network (2020, methodology): “Following the protocols of [38,37,41], we further com-”
Deep Learning for Single Image Super-Resolution: A Brief Review (2018, methodology): “We are grateful to the authors of [47], [84], [71], [61], [68], [121], [116], [114], [96], [92], [100] for kindly releasing their experimental results or codes, as well as to the three anonymous reviewers for their constructive criticism, which has significantly improved our manuscript.”
Unfolding the Alternating Optimization for Blind Super Resolution (2020, methodology): “Following [14, 37], the kernel is also reshaped and then reduced by principal component analysis (PCA).”
Blind image super-resolution with spatially variant degradations (2019, methodology): “In the non-blind setting, the kernel k is known and it is possible to have the synthesis process adapt to it [Zhang et al. 2018d].”
Multilevel and Multiscale Network for Single-Image Super-Resolution (2019, methodology): “To illustrate the effectiveness of our method, we compare the proposed method with other SR methods, including the Bicubic, SRCNN [10], VDSR [13], DRCN [26], MS-LapSRN [27], and SRMDNF [28] methods.”
Residual Dense Network for Image Restoration (2018, methodology): “Recently, deep convolutional neural network (CNN) has shown its excellent performance in different image restoration tasks, such as image super-resolution (SR) [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], image denoising (DN) [15], [16], [17], [18], [19], [20], and image compression r”
Progressive Fusion Video Super-Resolution Network via Exploiting Non-Local Spatio-Temporal Correlations (2019, methodology): “[4, 5], a lot of CNN-based SR methods [8, 24, 27, 33, 40] have been proposed.”
Image De-Raining Transformer (2022, neutral): “…used in various image processing tasks [17], [18], [31], [32], [33], [34], [35], such as de-noising [36], [37], [38], super-resolution [39], [40], [41], de-blurring [42], [43], [44], [45], de-hazing [19], [19], [46], [47], [48], [49], [50], low-light image enhancement [51], [52], [53], [54],…”
PIPAL: a Large-Scale Image Quality Assessment Dataset for Perceptual Image Restoration (2020, methodology): “g-based algorithms. In addition to Gaussian noise, we also include JPEG compression noise removal results. At last, we include the restoration results of the mixture image degradation. As revealed in [54,38], performing denoising and SR sequentially or jointly will bring new artifacts or dierent blu”
MADNet: A Fast and Lightweight Network for Single-Image Super Resolution (2020, methodology): “Speciﬁcally, MADNet has similar model size to those of DRCN [19], MemNet [38], and SRMDNF [54], while we achieve a better performance than all of them.”
Residual Feature Distillation Network for Lightweight Image Super-Resolution (2020, methodology): “We compare the proposed RFDN with various lightweight SR methods on ×2, ×3 and ×4 scales, including SRCNN [4], FSRCNN [5], VDSR [12], DRCN [13], LapSRN [14], DRRN [22], MemNet [23], IDN [11], SRMDNF [31], CARN [1] and IMDN [10].”
Blind Super-Resolution With Iterative Kernel Correction (2019, methodology): “SRMD [39] propose a stretching strategy to integrate non-image degradation information in a SR network.”
Unsupervised Degradation Representation Learning for Blind Super-Resolution (2021, methodology): “Speciﬁcally, degradation is ﬁrst used as an additional input in SRMD [42] to super-resolve LR images under different degradations.”
Cooperative Semantic Segmentation and Image Restoration in Adverse Environmental Conditions (2019, background): “[51] learn a single convolutional super-resolution network SRMD with high scalability for multiple degradations.”
Residual Dense Network for Image Super-Resolution (2018, methodology): “To tackle this inverse problem, plenty of image SR algorithms have been proposed, including interpolationbased [40], reconstruction-based [37], and learning-based methods [28, 29, 20, 2, 21, 8, 10, 31, 39].”