Image super-resolution using a Bayesian approach with a Gauss-Markov-Potts prior

Majdi Mansouri, A. Mohammad-Djafari

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Super-Resolution (SR) image reconstruction can be viewed as a three tasks process: registration or motion estimation, Point Spread Function (PSF) estimation and high resolution (HR) image reconstruction. In this paper, we propose a new method based on the Bayesian estimation with a Gauss-Markov-Potts prior model (MGMPP) where the main objective is to get a new High Resolution (HR) image from a set of severely blurred, noisy, rotated and shifted Low Resolution (LR) images. As a byproduct of our prior model, we obtain jointly a SR image and an optimal segmentation of it. The proposed algorithm is unsupervised. A comparison of the performances of the proposed method with some classical and recent SR methods are provided in simulation. The simulation results show that Bayesian estimation with a MGMPP algorithm, outperforms methods proposed by Marcel and al, Lucchese and Cortelazzo, Vanderwalle, and Keren.

Original languageEnglish
Title of host publicationProceedings of the 2009 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2009
Pages814-818
Number of pages5
Volume2
Publication statusPublished - 2009
Externally publishedYes
Event2009 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2009 - Las Vegas, NV, United States
Duration: 13 Jul 200916 Jul 2009

Other

Other2009 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2009
CountryUnited States
CityLas Vegas, NV
Period13/7/0916/7/09

Fingerprint

Image resolution
Image reconstruction
Potts model
Optical transfer function
Motion estimation
Byproducts

Keywords

  • Bayesian estimation
  • Motion estimation
  • Prior modeling
  • PSF estimation
  • Segmentation
  • Super Resolution

ASJC Scopus subject areas

  • Computer Graphics and Computer-Aided Design
  • Computer Vision and Pattern Recognition

Cite this

Mansouri, M., & Mohammad-Djafari, A. (2009). Image super-resolution using a Bayesian approach with a Gauss-Markov-Potts prior. In Proceedings of the 2009 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2009 (Vol. 2, pp. 814-818)

Image super-resolution using a Bayesian approach with a Gauss-Markov-Potts prior. / Mansouri, Majdi; Mohammad-Djafari, A.

Proceedings of the 2009 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2009. Vol. 2 2009. p. 814-818.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Mansouri, M & Mohammad-Djafari, A 2009, Image super-resolution using a Bayesian approach with a Gauss-Markov-Potts prior. in Proceedings of the 2009 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2009. vol. 2, pp. 814-818, 2009 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2009, Las Vegas, NV, United States, 13/7/09.
Mansouri M, Mohammad-Djafari A. Image super-resolution using a Bayesian approach with a Gauss-Markov-Potts prior. In Proceedings of the 2009 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2009. Vol. 2. 2009. p. 814-818
Mansouri, Majdi ; Mohammad-Djafari, A. / Image super-resolution using a Bayesian approach with a Gauss-Markov-Potts prior. Proceedings of the 2009 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2009. Vol. 2 2009. pp. 814-818
@inproceedings{388381a228fb46a486270bff66dab0ae,
title = "Image super-resolution using a Bayesian approach with a Gauss-Markov-Potts prior",
abstract = "Super-Resolution (SR) image reconstruction can be viewed as a three tasks process: registration or motion estimation, Point Spread Function (PSF) estimation and high resolution (HR) image reconstruction. In this paper, we propose a new method based on the Bayesian estimation with a Gauss-Markov-Potts prior model (MGMPP) where the main objective is to get a new High Resolution (HR) image from a set of severely blurred, noisy, rotated and shifted Low Resolution (LR) images. As a byproduct of our prior model, we obtain jointly a SR image and an optimal segmentation of it. The proposed algorithm is unsupervised. A comparison of the performances of the proposed method with some classical and recent SR methods are provided in simulation. The simulation results show that Bayesian estimation with a MGMPP algorithm, outperforms methods proposed by Marcel and al, Lucchese and Cortelazzo, Vanderwalle, and Keren.",
keywords = "Bayesian estimation, Motion estimation, Prior modeling, PSF estimation, Segmentation, Super Resolution",
author = "Majdi Mansouri and A. Mohammad-Djafari",
year = "2009",
language = "English",
isbn = "9781601321190",
volume = "2",
pages = "814--818",
booktitle = "Proceedings of the 2009 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2009",

}

TY - GEN

T1 - Image super-resolution using a Bayesian approach with a Gauss-Markov-Potts prior

AU - Mansouri, Majdi

AU - Mohammad-Djafari, A.

PY - 2009

Y1 - 2009

N2 - Super-Resolution (SR) image reconstruction can be viewed as a three tasks process: registration or motion estimation, Point Spread Function (PSF) estimation and high resolution (HR) image reconstruction. In this paper, we propose a new method based on the Bayesian estimation with a Gauss-Markov-Potts prior model (MGMPP) where the main objective is to get a new High Resolution (HR) image from a set of severely blurred, noisy, rotated and shifted Low Resolution (LR) images. As a byproduct of our prior model, we obtain jointly a SR image and an optimal segmentation of it. The proposed algorithm is unsupervised. A comparison of the performances of the proposed method with some classical and recent SR methods are provided in simulation. The simulation results show that Bayesian estimation with a MGMPP algorithm, outperforms methods proposed by Marcel and al, Lucchese and Cortelazzo, Vanderwalle, and Keren.

AB - Super-Resolution (SR) image reconstruction can be viewed as a three tasks process: registration or motion estimation, Point Spread Function (PSF) estimation and high resolution (HR) image reconstruction. In this paper, we propose a new method based on the Bayesian estimation with a Gauss-Markov-Potts prior model (MGMPP) where the main objective is to get a new High Resolution (HR) image from a set of severely blurred, noisy, rotated and shifted Low Resolution (LR) images. As a byproduct of our prior model, we obtain jointly a SR image and an optimal segmentation of it. The proposed algorithm is unsupervised. A comparison of the performances of the proposed method with some classical and recent SR methods are provided in simulation. The simulation results show that Bayesian estimation with a MGMPP algorithm, outperforms methods proposed by Marcel and al, Lucchese and Cortelazzo, Vanderwalle, and Keren.

KW - Bayesian estimation

KW - Motion estimation

KW - Prior modeling

KW - PSF estimation

KW - Segmentation

KW - Super Resolution

UR - http://www.scopus.com/inward/record.url?scp=84864919153&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84864919153&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9781601321190

VL - 2

SP - 814

EP - 818

BT - Proceedings of the 2009 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2009

ER -