Experimental tests of image fusion for night vision

Abstract

Image fusion techniques have begun to play a very important role in night vision systems. In recent years, various image fusion algorithms have been developed to perform the task. However, very few comprehensive studies have been conducted to evaluate the performance of fusion methods for night vision applications. In this paper we focus on fusion algorithms especially for the night vision application and employ experimental testing to compare their performance. To judge the performance of image fusion algorithms, we investigate both subjective and objective evaluation measures. Human evaluations of the fusion results are presented. Furthermore, to evaluate image fusion algorithms objectively, we studied various image quality measures, which include some standard quality metrics and other newly developed methods. Extensive performance evaluation experiments show that observers generally prefer the SiDWT and Laplacian pyramid fusion scheme for the considered test images. Further the objective quality measure results show that edge based quality metrics follows human evaluations much closer than the other methods in most of the cases we considered.