Limits on super-resolution and how to break them

By Simon Baker, Takeo Kanade. 2000

1,669 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

The Perception-Distortion Tradeoff (2017, background): “Several past works attempted to answer the question: What is the minimal attainable distortion Dmin in various restoration tasks? [49], [50], [51], [52], [53].”
PULSE: Self-Supervised Photo Upsampling via Latent Space Exploration of Generative Models (2020, background): “The problem of recovering the true HR image depicted by an LR input, as opposed to generating a set of potential such HR images, is inherently ill-posed, as the size of the total set of these images grows exponentially with the scale factor [2].”
A two-step approach to hallucinating faces: global parametric model and local nonparametric model (2001, background): “They also discuss the limits on super resolution and how to break them in their method [2].”
Image Mosaicing and Super-resolution (2004, methodology): “Some authors have adopted a simpler motion model, such as pure translation [10, 94, 104, 106, 114, 152, 164, 168], Euclidean [34, 54, 86] or affine [11, 87].”
Fast image/video upsampling (2008, positive): “Typical approaches [Baker and Kanade 2000; Tipping and Bishop 2002] formulate the reconstruction process with an image formation model.”
Image super-resolution using gradient profile prior (2008, methodology): “The learning based methods [10, 9, 26, 5, 28, 2, 7, 20, 31] “hallucinate” high frequency details from a training set of HR/LR image pairs.”
A Comprehensive Survey to Face Hallucination (2013, methodology): “Reconstruction-based methods estimate a highresolution image from a sequence of blurred and downsampled low-resolution images (Elad and Feuer, 1997, 1999; Hardie et al, 1997) and there are inherent limitations in relation to increasing the magnification factor (Baker and Kanade, 2002).”
Iterative kernel principal component analysis for image modeling (2005, neutral): “[3]).”
Natural image denoising: Optimality and inherent bounds (2011, background): “In fact, in the same paper, Baker and Kanade [1] show that their super-resolution limits can be broken using a face class prior.”
Deep Networks for Image Super-Resolution with Sparse Prior (2015, background): “With the known variables in LR images greatly outnumbered by the unknowns in HR images, SR is a highly ill-posed problem and the current techniques are far from being satisfactory for many real applications [2, 21].”
Super-resolution through neighbor embedding (2004, background): “, [2, 6, 8, 22]), the focus of this paper is on generating a high-resolution image from a single low-resolution image, with the help of a set of one or more training images from scenes of the same or different types.”
3D Object Modeling and Recognition Using Local Affine-Invariant Image Descriptors and Multi-View Spatial Constraints (2006, positive): “2001; Baker and Kanade, 2002 ), provided the patches satisfy our assumption of planarity and that they are well registered.”
Image up-sampling using total-variation regularization with a new observation model (2005, background): “Baker and Kanade [42] involved in their analysis a general form for the optical blurring effect convolved with a rectangular function that models a CCD camera used to acquire the LR image.”
Motion-based motion deblurring (2004, background): “Also, note that the secondary detector is used only as a motion sensor; it has low resolution and high gain and is not suitable for superresolution purposes [1].”
Image hallucination with primal sketch priors (2003, background): “Unlike “face hallucination” [3], however, our approach works for generic images.”