Joint parameter estimation and restoration using MRF models and homotopy continuation method

Abstract

This paper presents a joint strategy for parameter estimation of Markov Random Field (MRF) model and image restoration. The proposed scheme is an unsupervised one in the sense that no a priori knowledge of the actual image is assumed. The technique of homotopy continuation method is employed to estimate the model parameters. The model considered involves line fields and is tested on real images. Simulation results are presented for gray level images.