• Corpus ID: 218972161

Image Restoration from Parametric Transformations using Generative Models

@article{Basioti2020ImageRF,
  title={Image Restoration from Parametric Transformations using Generative Models},
  author={Kalliopi Basioti and George V. Moustakides},
  journal={ArXiv},
  year={2020},
  volume={abs/2005.14036}
}
When images are statistically described by a generative model we can use this information to develop optimum techniques for various image restoration problems as inpainting, super-resolution, image coloring, generative model inversion, etc. With the help of the generative model it is possible to formulate, in a natural way, these restoration problems as Statistical estimation problems. Our approach, by combining maximum a-posteriori probability with maximum likelihood estimation, is capable of… 

Figures and Tables from this paper

Image De-Quantization Using Generative Models as Priors
TLDR
The goal in this work is to develop a de-quantization mechanism through a rigorous mathematical analysis which is based on the classical statistical estimation theory and includes generative modeling of the ideal image as a suitable prior information.
An Image Restoration Architecture using Abstract Features and Generative Models
TLDR
An image restoration architecture that enables us to address different types of corruption, regardless of type, amount and location is proposed and has been compared with both traditional approaches like BM3D and with modern deep models like IRCNN and NCSR.
Model Adaptation for Inverse Problems in Imaging
TLDR
This work proposes two novel procedures that adapt the network to a change in the forward model, even without full knowledge of the change, and shows these simple model adaptation approaches achieve empirical success in a variety of inverse problems, including deblurring, super-resolution, and undersampled image reconstruction in magnetic resonance imaging.

References

SHOWING 1-10 OF 53 REFERENCES
Predictive modeling of the seismic cycle of the Greater San Francisco Bay Region
The seismic cycle for the San Francisco Bay region is synthesized by a model combining the pre-and post-1906 seismic histories. The long-term acceleration of seismic release (seismic moment, Benioff
Spatial indoor radon distribution in Mexico City.
Electron transport in Neurospora mitochondria. Studies on wild type and poky.
TLDR
Two lines of evidence suggest that, in poky, the cytochrome chain and the alternate oxidase compete for reducing equivalents on the oxygen side of the dehydrogenases, and thus form a branched electron transport system.
Compressed Sensing with Invertible Generative Models and Dependent Noise
TLDR
This work studies image inverse problems with invertible generative priors, specifically normalizing flow models, and establishes theoretical recovery guarantees for denoising and compressed sensing under its framework.
Random Processes for Engineers
  • B. Hajek
  • Mathematics, Computer Science
  • 2015
TLDR
This work presents a meta-analyses of random processes in linear systems and spectral analysis using countable-state Markov models and its applications to even numbered problems.
Generative Adversarial Source Separation
  • Y. C. Sübakan, P. Smaragdis
  • Computer Science
    2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
  • 2018
TLDR
It is shown on a speech source separation experiment that, a multilayer perceptron trained with a Wasserstein-GAN formulation outperforms NMF, auto-encoders trained with maximum likelihood, and variational auto- Encoders in terms of source to distortion ratio.
Your Local GAN: Designing Two Dimensional Local Attention Mechanisms for Generative Models
TLDR
This work introduces a new local sparse attention layer that preserves two-dimensional geometry and locality in SAGAN and presents a novel way to invert Generative Adversarial Networks with attention.
Compressed Sensing using Generative Models
TLDR
This work shows how to achieve guarantees similar to standard compressed sensing but without employing sparsity at all, and proves that, if G is L-Lipschitz, then roughly O(k log L) random Gaussian measurements suffice for an l2/l2 recovery guarantee.
Deep Learning Face Attributes in the Wild
TLDR
A novel deep learning framework for attribute prediction in the wild that cascades two CNNs, LNet and ANet, which are fine-tuned jointly with attribute tags, but pre-trained differently.
Imagination Can Create False Autobiographical Memories
Previous studies have shown that imagining an event can alter autobiographical beliefs. The current study examined whether it can also create false memories. One group of participants imagined a
...
...