On the Role of Sparse and Redundant Representations in Image Processing

@article{Elad2010OnTR,
  title={On the Role of Sparse and Redundant Representations in Image Processing},
  author={Michael Elad and M{\'a}rio A. T. Figueiredo and Yi Ma},
  journal={Proceedings of the IEEE},
  year={2010},
  volume={98},
  pages={972-982}
}
Much of the progress made in image processing in the past decades can be attributed to better modeling of image content and a wise deployment of these models in relevant applications. This path of models spans from the simple l2-norm smoothness through robust, thus edge preserving, measures of smoothness (e.g. total variation), and until the very recent models that employ sparse and redundant representations. In this paper, we review the role of this recent model in image processing, its… 

Figures from this paper

Image super-resolution via two coupled dictionaries and sparse representation
TLDR
A novel technique is proposed that can generate a SR image from a single LR input image via convolution operation and the principal component analysis (PCA) is used to reduce information redundancy after features extraction step.
A Review of Adaptive Image Representations
  • G. Peyré
  • Computer Science
    IEEE Journal of Selected Topics in Signal Processing
  • 2011
TLDR
This paper reviews these emerging technics and shows the interplay between sparse and non-local regularizations.
Color Sparse Representations for Image Processing: Review, Models, and Prospects
TLDR
It is shown that the scalar quaternionic linear model is equivalent to constrained matrix-based color filtering, which highlights the filtering implicitly applied through this model.
Multi-view object recognition based on sparse representation
TLDR
A novel framework called metasample-based supervised dictionary learning for multi-view object recognition exploiting the sparse property of intrinsic information was proposed and Experimental results demonstrate that the proposed algorithm exhibits better performance than the recent state-of-the-art methods.
Learning Sparse Data Models via Geometric Optimization with Applications to Image Processing
TLDR
This thesis investigates the problem of learning sparse data models and their applications to image processing, regarding both the synthesis and the analysis point of view, and proposes an extension of the analysis model that is able to model dependencies of different signal modalities.
Comparative analysis of inpainting techniques based on sparse models and isophote comparison
TLDR
This work compares the results obtained with sparse modelling against those obtained with two other techniques, the first one based on bilinear interpolation and the second one, called Isophote continuation, which initially identifies the area to be reconstructed, then from the adjacent neighbours creates new layers within the region to be reconstruction and repeats the process until the areas to be restored is completely filled.
Analysis Operator Learning and its Application to Image Reconstruction
TLDR
This paper presents an algorithm for learning an analysis operator from training images based on lp-norm minimization on the set of full rank matrices with normalized columns, and carefully introduces the employed conjugate gradient method on manifolds.
Context-Aware Sparse Decomposition for Image Denoising and Super-Resolution
TLDR
The contextual information of local patches is utilized as context-aware sparsity prior to enhance the performance of sparsity-based restoration method and a unified framework based on the Markov random fields model is proposed to tune the local prior into a global one to deal with arbitrary size images.
Nonlocally Centralized Sparse Representation for Image Restoration
TLDR
The so-called nonlocally centralized sparse representation (NCSR) model is as simple as the standard sparse representation model, and the extensive experiments validate the generality and state-of-the-art performance of the proposed NCSR algorithm.
Image Deblurring and Super-Resolution by Adaptive Sparse Domain Selection and Adaptive Regularization
TLDR
Extensive experiments on image deblurring and super-resolution validate that by using adaptive sparse domain selection and adaptive regularization, the proposed method achieves much better results than many state-of-the-art algorithms in terms of both PSNR and visual perception.
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 107 REFERENCES
Sparse Representation for Color Image Restoration
TLDR
This work puts forward ways for handling nonhomogeneous noise and missing information, paving the way to state-of-the-art results in applications such as color image denoising, demosaicing, and inpainting, as demonstrated in this paper.
Multiscale Hybrid Linear Models for Lossy Image Representation
TLDR
The careful and extensive experimental results show that this new model gives more compact representations for a wide variety of natural images under a wide range of signal-to-noise ratios than many existing methods, including wavelets.
Image super-resolution as sparse representation of raw image patches
TLDR
It is shown that a small set of randomly chosen raw patches from training images of similar statistical nature to the input image generally serve as a good dictionary, in the sense that the computed representation is sparse and the recovered high-resolution image is competitive or even superior in quality to images produced by other SR methods.
Image Denoising with Shrinkage and Redundant Representations
TLDR
It is shown that simple shrinkage could be interpreted as the first iteration of an algorithm that solves the basis pursuit denoising (BPDN) problem, which leads to a novel iterative shrinkage algorithm that can be considered as an effective pursuit method.
A multiscale hybrid linear model for lossy image representation
TLDR
This paper introduces a simple and efficient representation for natural images that gives more compact representations for a wide variety of natural images under a wide range of signal-to-noise ratio than many existing methods, including wavelets.
Image Denoising Via Sparse and Redundant Representations Over Learned Dictionaries
TLDR
This work addresses the image denoising problem, where zero-mean white and homogeneous Gaussian additive noise is to be removed from a given image, and uses the K-SVD algorithm to obtain a dictionary that describes the image content effectively.
Recent Developments in Total Variation Image Restoration
TLDR
There has been a resurgence of interest and exciting new developments in total variation minimizing models, some extending the applicabilities to inpainting, blind deconvolution and vector-valued images, while others offer improvements in better preservation of contrast, geometry and textures.
Nonlinear approximation based image recovery using adaptive sparse reconstructions and iterated denoising-part I: theory
  • O. Guleryuz
  • Computer Science
    IEEE Transactions on Image Processing
  • 2006
TLDR
The robust estimation of missing regions in images and video using adaptive, sparse reconstructions using constructed estimators and how these estimators relate to the utilized transform and its sparsity over regions of interest is shown.
Nonlinear approximation based image recovery using adaptive sparse reconstructions
  • O. Guleryuz
  • Mathematics, Computer Science
    Proceedings 2003 International Conference on Image Processing (Cat. No.03CH37429)
  • 2003
TLDR
The robust estimation of missing regions in images and video using adaptive, sparse reconstructions is studied and it is shown that the region types the authors can effectively estimate in a mean squared error sense are those for which the given transform provides a close approximation using nonlinear approximation.
...
1
2
3
4
5
...