Nonparametric Blind Super-resolution

By Tomer Michaeli, Michal Irani. 2013

323 citations recorded by the source corpus.

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): “As mentioned in the introduction, existing DNNs-based SISR methods are generally based on bicubic downsam-pling [23, 44] and traditional degradations [26, 37, 54, 59, 60], or some simple variants [11, 41, 53, 56, 58].”
Real-ESRGAN: Training Real-World Blind Super-Resolution with Pure Synthetic Data (2021, background): “Blind super-resolution [34, 2, 54], on the contrary, aims to restore low-resolution images suffering from unknown and complex degradations.”
Jointly Optimized Regressors for Image Super‐resolution (2015, background): “Recent approaches try to estimate appropriate blur kernels instead ﬁnding good image priors [EGA ∗ 13,MI13].”
A+: Adjusted Anchored Neighborhood Regression for Fast Super-Resolution (2014, background): “Some recent approaches try to shift the focus from finding good image priors to finding an appropriate blur kernel [6, 7].”
Unsupervised Image Super-Resolution Using Cycle-in-Cycle Generative Adversarial Networks (2018, methodology): “Blind SR [19,29] follow the same assumption, only with unknown SH .”
Blind image super-resolution with spatially variant degradations (2019, methodology): “In the blind setting, Shocher et al. [2018] train a network specifically for each image after recovering the blur kernel using patch repetition assumption [Michaeli and Irani 2013].”
Deep Learning for Image Super-Resolution: A Survey (2019, methodology): “However, the previous studies based on back-projection are mostly not deep learning based and involve some unlearnable operations [84], [85].”
Deep Learning for Single Image Super-Resolution: A Brief Review (2018, methodology): “Notably, when combined with the kernel estimation algorithms mentioned in [102], ZSSR performs quite well with the unknown degradation kernels.”
Scaling Up to Excellence: Practicing Model Scaling for Photo-Realistic Image Restoration In the Wild (2024, background): “However, these methods are often based on speciﬁc degradation assumptions [25, 50, 56] and therefore lack generalization ability to other degradations [29, 52, 94].”
Single-Image Super-Resolution: A Benchmark (2014, background): “Although it is widely accepted that the blur kernel significantly affects the performance of SISR algorithms [21,6], the proper values of blur kernel width have not been thoroughly investigated.”
SwinIR: Image Restoration Using Swin Transformer (2021, methodology): “Compared to traditional image restoration methods [28, 72, 73, 62, 32] which are generally model-based, learningbased methods, especially CNN-based methods, have become more popular due to their impressive performance.”
Flow-based Kernel Prior with Application to Blind Super-Resolution (2021, methodology): “Prior to the deep learning era, traditional kernel estimation methods typically utilize prior information of image patches or edges [2, 21, 35, 41, 46].”
Seven Ways to Improve Example-Based Single Image Super Resolution (2015, background): “Knowing the degradation operators is a must for the IBP approaches and therefore they need to be estimated [21].”
Blind Super-Resolution With Iterative Kernel Correction (2019, background): “Note that the formulation of blind SR in this paper is different with the previous works [23, 37] .”
Restormer: Efficient Transformer for High-Resolution Image Restoration (2021, background): “In recent years, data-driven CNN architectures [7, 18, 92, 93, 105, 107] have been shown to outperform conventional restoration approaches [26,36,53,75].”
Zero-Shot Super-Resolution Using Deep Internal Learning (2017, methodology): “For this mode we used [14] to evaluate the kernel directly from the test image and fed it to ZSSR.”
Blind Super-Resolution Kernel Estimation using an Internal-GAN (2019, methodology): “We compare our kernel estimation to Michaeli & Irani [24] which, to the best of our knowledge, is the leading method for SR-kernel estimation from a single image.”
Unsupervised Degradation Representation Learning for Blind Super-Resolution (2021, methodology): “Existing degradation estimation methods [28, 3, 12] aim at estimating the degradation (usually the blur kernel) at pixel level.”