Abstract



Mutual Information for the Evaluation of Various Image Fusion Methods

Wang Zhijun
Dept of Mathematics and Informatics
Faculty of Sciences, Université de Sherbrooke
Québec, Canada
Email: zhi.jun.wang@usherbrooke.ca

Djemel Ziou


Abstract
Correlation is one of the most common similarity metrics used in image fusion to judge how the spectral characteristics are preserved quantitatively. However, it is not sensitive enough to represent the slight difference of the spectral characteristics in the fused results. Mutual Information (MI) measures the amount of information that one image contains about another image by looking at their intensity distribution and usually produce more sensitive results than correlation. Experiments show that the MI is better similarity metric than correlation in image fusion.