Fast Direct Super-Resolution by Simple Functions

By Chih-Yuan Yang, Ming-Hsuan Yang. IEEE International Conference on Computer Vision, 2013

337 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/028496484cd84d778f736798463de1e4b8df3c49

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

Image Super Resolution Based on Fusing Multiple Convolution Neural Networks (2017, methodology): “Various learning algorithms have been used, including nearest neighbor approaches [22], manifold learning [7], dictionary learning [11][10], and locally linear regression [34].”
Single image super-resolution from transformed self-exemplars (2015, methodology): “…global mapping over the entire dataset, several methods alleviate the complexity of data modeling by partitioning or pre-clustering the external training database, so that relatively simpler prediction functions could be used for performing the LR-HR mapping in each training cluster [38] [33] [34].”
Naive Bayes Super-Resolution Forest (2015, methodology): “This produces a better sampling of the coarse manifold and also provides a unification of [28] and [23].”
Enhanced Deep Residual Networks for Single Image Super-Resolution (2017, background): “[30] introduced another approach that clusters the patch spaces and learns the corresponding functions.”
Learning a Deep Convolutional Network for Image Super-Resolution (2014, methodology): “A similar trend is observed when we used SSIM [22,21] as the performance metric, the results of which could be found in the supplementary file.”
Fast and Accurate Image Super-Resolution with Deep Laplacian Pyramid Networks (2017, methodology): “Instead of directly modeling the complex patch space over the entire database, recent methods partition the image set by K-means [4], sparse dictionary [3], [28] or random forest [5], and learn”
End-to-End Image Super-Resolution via Deep and Shallow Convolutional Networks (2016, background): “For instance, Yang et al. [6] perform SR by training the LR and HR dictionaries jointly with the constraint that LR patches and its corresponding HR counterparts share the same sparse representation.”
Accelerating the Super-Resolution Convolutional Neural Network (2016, methodology): “Recent SR algorithms are mostly learning-based (or patch-based) methods [1,2,3,4,5,6,7,8] that learn a mapping between the LR and HR image spaces.”
A+: Adjusted Anchored Neighborhood Regression for Fast Super-Resolution (2014, methodology): “This is what Yang and Yang [12] propose.”
Ultra-Resolving Face Images by Discriminative Generative Networks (2016, methodology): “Generic super-resolution: In general, generic single image super-resolution methods have three types: interpolation based methods, image statistics based methods [20,21] and example (patch)-based methods [22,23,24,7,25,26].”
Learning to Super-Resolve Blurry Face and Text Images (2017, methodology): “Regression-based methods typically learn the priors from patches [48, 40].”
Wavelet-SRNet: A Wavelet-Based CNN for Multi-scale Face Super Resolution (2017, background): “In general, image super-resolution methods can be divided into three types: interpolation-based, statisticsbased [26, 31, 32] and learning-based methods [3, 9, 24].”
Single-Image Super-Resolution: A Benchmark (2014, methodology): “The formation is compatible with most SR methods [16,11,17,30,46,43,36] where the reconstructed images are well aligned with the ground truth images.”
Convolutional Sparse Coding for Image Super-Resolution (2015, methodology): “For mapping-based methods, mapping function between the LR and HR images is directly learned using the LR/HR patch pairs to implicitly incorporate prior knowledge [5, 6, 26, 11].”
SRFeat: Single Image Super-Resolution with Feature Discrimination (2018, background): “neighborhood embedding [7, 45, 46, 36], sparse coding [52, 54, 51, 16], and local mapping function regression [15, 50, 38].”