ABSTRACT

Image fusion integrates information from multiple input images acquired by different imaging modalities or by same modality operated in different conditions. Image fusion assessment includes algorithm parameters, subjective scores, and objective metrics. This chapter describes the role and implementation of pixel-level fusion. It presents the state-of-the-art of objective metrics and describes two trials on subjective assessment. The chapter explores the role of natural imagery interpretability rating scales (NIIRSs) on fusion performance assessment. It then illustrates the procedure to apply statistical approaches to analyze both subjective and objective fusion assessment data. The chapter provides the experimental results on the assessment of fused night vision data sets. It also highlights future research perspectives on cognitive assessment. Pixel-level fusion generates a composite from the input images, integrates the complementary information into the fused result, and reduces the uncertainty with redundant information. There are numerous algorithms proposed for pixel-level image fusion. One implementation is in the multiresolution transform domain known as multiresolution analysis (MRA).