A General Multichannel Image Restoration Method Using Compound Models

In this paper we present a multichannel image restoration method using Compound Gauss Markov Random Field (CGMRF) models. Information regarding the objects present in the scene is shared via the line process in the CGMRF. Two new iterative algorithms to estimate the underlying multichannel image are presented, which can be considered as extensions of the classical simulated annealing and ICM methods. Experimental results demonstrate the effectiveness of the proposed approach.

• Type: Conference
• Year: 2002
• Where: ICPR
• Authors: Rafael Molina, Javier Mateos, Aggelos K. Katsaggelos, Miguel Vega
• Keywords: Classical Simulated Annealing, Compound Gauss Markov, Computer Vision, ICPR 2002, Multichannel Image Restoration