Image and video upscaling from local self-examples

By Gilad Freedman, Raanan Fattal. TOGS, 2011

689 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/95064fdabc04e1621ec1e46ff85f74163ba2f766

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

Citations in the corpus

Learning Enriched Features for Fast Image Restoration and Enhancement (2022, background): “Prior to the deep-learning era, numerous super-resolution (SR) algorithms have been proposed based on the sampling theory [43], [44], edge-guided interpolation [45], [46], natural image priors [47], [48], patch-exemplars [49], [50] and sparse representations [51], [52].”
Deep Learning for Image Super-Resolution: A Survey (2019, methodology): “In literature, a variety of classical SR methods have been proposed, including prediction-based methods [10], [11], [12], edge-based methods [13], [14], statistical methods [15], [16], patch-based methods [13], [17], [18], [19] and sparse representation methods [20], [21], etc.”
Learning Enriched Features for Real Image Restoration and Enhancement (2020, methodology): “Prior to the deep-learning era, numerous SR algorithms have been proposed based on the sampling theory [53,51], edge-guided interpolation [4,64,120], natural image priors [56,109], patch-exemplars [14,32] and sparse representations [113,112].”
Non-Local Recurrent Network for Image Restoration (2018, neutral): “Self-example based image SR approaches [14, 12] exploit the local self-similarity assumption, and extract LR-HR exemplar pairs merely from the low-resolution image across different scales to predict the high-resolution image.”
Structure-Preserving Super Resolution With Gradient Guidance (2020, methodology): “Here we review SISR methods [7, 10, 12, 13, 14, 19, 22, 25, 38, 44, 46, 47] which can be classified into two categories: PSNR-oriented methods and perceptual-driven ones.”
Low-Complexity Single-Image Super-Resolution based on Nonnegative Neighbor Embedding (2012, methodology): “In local learning methods, the mapping from a LR patch to a HR patch is inferred by using Support Vector Regression [11], linear or Kernel Ridge Regression [13].”
Balanced Two-Stage Residual Networks for Image Super-Resolution (2017, methodology): “Given the origin of example pairs, these methods can be further categorized into three classes: self-example based [9, 7], external-example based methods [42, 34] and the joint of them [39].”
A Statistical Prediction Model Based on Sparse Representations for Single Image Super-Resolution (2014, background): “Models that are commonly exploited in single image SR methods include image smoothness and geometric regularity of image structures [3]), gradient profile priors [4], self-similarity of image patches within and across different scales in the same image [6], [7], [24], and sparsity – either in the wa”
Toward Real-World Single Image Super-Resolution: A New Benchmark and a New Model (2019, background): “age prior information from HR and/or LR exemplar images [16, 14, 57, 20, 15, 8, 25, 58, 12, 21, 47, 42] plays an indispensable role in recovering details from an LR image.”
Frame-Recurrent Video Super-Resolution (2018, methodology): “Super-resolution is a classic ill-posed inverse problem with approaches ranging from simple interpolation methods such as Bilinear, Bicubic and Lanczos [8] to example-based super-resolution [11, 12, 37, 39], dictionary learning [30, 40], and self-similarity approaches [15, 38].”
Deep Networks for Image Super-Resolution with Sparse Prior (2015, methodology): “9 shows the SR results on the “chip” image compared among more methods including the self-example based method (SE) [12] and the deep network cascade (DNC) [6].”
End-to-End Image Super-Resolution via Deep and Shallow Convolutional Networks (2016, background): “Image SR can be generally classiﬁed into three categories, i.e. , interpolation-based [25], [26], reconstruction based [27]– [29], and learning-based methods [2], [3], [13].”
Nonparametric Blind Super-resolution (2013, background): “Such cross-scale patch recurrence has been shown to provide a strong cue for SR from a single image [8, 6].”
Learning Texture Transformer Network for Image Super-Resolution (2020, background): “Recent progress has been made by reference-based image super-resolution (RefSR), which transfers HR textures from a given Ref image to produce visually pleasing results [5, 6, 26, 29, 36].”
Accurate Blur Models vs. Image Priors in Single Image Super-resolution (2013, background): “Developing sophisticated image priors has been the focus of much single image SR research in the past decade, with many significant successes [7, 20, 23, 8, 6, 9, 21, 27, 25, 5, 19, 22, 1, 8, 13].”
Single image super-resolution from transformed self-exemplars (2015, methodology): “Freedman and Fattal [13] showed that self-similar patches can often be found in limited spatial neighborhoods, thereby gaining computational speed-up.”
Image Super-Resolution Using Deep Convolutional Networks (2014, methodology): “These methods either exploit internal similarities of the same image [5], [13], [16], [19], [47], or learn mapping functions from external low- and high-resolution exemplar pairs [2], [4], [6], [15], [23], [25], [37], [41], [42], [47], [48], [50], [51].”
Image Super-Resolution by Neural Texture Transfer (2019, methodology): “Following the self-example matching strategy [7], we first apply bicubic up-sampling on I to get an upscaled LR image ILR↑ that has the same spatial size as I.”