Fast and robust multiframe super resolution

By Sina Farsiu, M. Dirk Robinson, Michael Elad, Peyman Milanfar. 2004

2,130 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

Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network (2016, background): “Here we will focus on single image super-resolution (SISR) and will not further discuss approaches that recover HR images from multiple images, such as object images acquired from varying view points or temporal sequences of image frames [4, 11].”
Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network (2016, neutral): “These can be categorised as multi-image SR methods [1, 11] and exploit explicit redundancy by constraining the ill-posed problem with additional information and attempting to invert the downsampling process.”
Deep Learning Techniques for Inverse Problems in Imaging (2020, background): “Superresolution [60, 61] A = SB S is a subsampling operator (identity matrix with missing rows) and B is a blurring operator cooresponding to convolution with a blur kernel Inpainting [62] A = S S is a diagonal matrix where Si,i = 1 for the pixels that are sampled and Si,i = 0 for the pixels that ar”
A multi-frame image super-resolution method (2010, methodology): “Regularization-based approaches [9–11], learning-based SR method [14,18] , space-time SR method [15] and color image SR method [22,23] have been proposed.”
KinectFusion: real-time 3D reconstruction and interaction using a moving depth camera (2011, background): “This creates an effect similar to image superresolution [6] – adding greater detail than appears visible in the raw signal (see Figure 3).”
Image Super-Resolution Via Sparse Representation (2010, neutral): “Various regulari zation methods have been proposed to further stabilize the inversi on this ill-posed problem, such as [14], [11], [25].”
Frame-Recurrent Video Super-Resolution (2018, background): “Classical video and multi-frame super-resolution methods are generally formulated as optimization problems that are computationally very expensive to solve [2, 10, 25, 35].”
Progressive Fusion Video Super-Resolution Network via Exploiting Non-Local Spatio-Temporal Correlations (2019, methodology): “Most traditional video SR methods [2, 7, 20, 22] conduct pixel-level motion and blur kernel estimation based on image priors, trying to solve a complex optimization problem [23].”
Deep Video Super-Resolution Network Using Dynamic Upsampling Filters Without Explicit Motion Compensation (2018, methodology): “Traditional VSR (or multi-image super-resolution) algorithms take multiple LR frames as inputs and output HR frames by taking into account subpixel motions between the neighboring LR frames [28, 4, 23, 26].”
A super-resolution reconstruction algorithm for surveillance images (2010, methodology): “Therefore, to ensure fairness, each method was repeated several times with different regularization parameters and the best result was chosen as the output of the method [30].”
PhaseStain: the digital staining of label-free quantitative phase microscopy images using deep learning (2018, methodology): “To synthesize a high-resolution hologram (with a pixel size of ~0.37 μ m) using only the G1 channel of the Bayer pattern (R, G1, G2, and B), a shift-and-add based PSR algorithm was applied 42,44 .”
NTIRE 2017 Challenge on Single Image Super-Resolution: Dataset and Study (2017, background): “Single image super-resolution as well as image restoration research literature spans over decades [36, 20, 4, 13, 16, 3, 15, 14, 6, 32, 54, 30, 17, 23, 12, 47, 48, 10, 21].”