MAP-MRF super-resolution image reconstruction using maximum pseudo-likelihood parameter estimation

Authors:
Ana L. D. Martins;Alexandre L. M. Levada;Murillo R. P. Homem;Nelson D. A. Mascarenhas
Affiliations:
Universidade Federal de São Carlos, Departamento de Computação, São Carlos, SP, Brazil;Universidade de São Paulo, Instituto de Física de São Carlos, São Carlos, SP, Brazil;Universidade Federal de São Carlos, Departamento de Computação, São Carlos, SP, Brazil;Universidade Federal de São Carlos, Departamento de Computação, São Carlos, SP, Brazil
Venue:
ICIP'09 Proceedings of the 16th IEEE international conference on Image processing
Year:
2009

Citing 4
Cited 2

Markov random field modeling in image analysis

Markov random field modeling in image analysis
Limits on Super-Resolution and How to Break Them

IEEE Transactions on Pattern Analysis and Machine Intelligence
Super-resolution image reconstruction using the generalized isotropic multi-level logistic model

Proceedings of the 2009 ACM symposium on Applied Computing
Maximum pseudo likelihood estimation in network tomography

IEEE Transactions on Signal Processing

A novel MAP-MRF approach for multispectral image contextual classification using combination of suboptimal iterative algorithms

Pattern Recognition Letters
Spatio-temporal resolution enhancement of vocal tract MRI sequences based on image registration

Integrated Computer-Aided Engineering

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we address the parameter estimation of a Super-Resolution Image Reconstruction approach following a Maximum a Posteriori Probability (MAP) algorithm. The Generalized Isotropic Multi-Level Logistic (GIMLL) Markov Random Field (MRF) model is considered for the high-resolution image characterization. In most applications, MRF model parameters are still chosen by a trial-and-error procedure through simple manual adjustments. In order to overcome this problem we propose a novel approach based on interval parameter estimation using both the Maximum Pseudo-Likelihood (MPL) technique and an approximation of the asymptotic variance of this estimator. To evaluate the capability of the proposed estimator we used a Markov Chain Monte Carlo algorithm to generate GIMLL model outcomes. The differences between the real parameters and the proposed MPL estimators are not significant. Moreover, the Normalized Mean Square Error (NMSE) of the high-resolution estimations indicate the effectiveness of our approach and the importance of an accurate estimation procedure.