An image fusion process should preserve all useful patterns from the source images while minimizing artifacts that could interfere with subsequent analyses or distract human observers. Given that it is nearly impossible to fuse images without introducing some form of distortion, measurements are necessary to present a fused image quality (IQ) for user analysis.
Image-quality measurement is as important as image fusion methods to guide developments for engineers, support learning methods for machines, and enhance trust with users. This chapter focuses on objective evaluation using quantitative metrics, whereas subjective evaluation will be discussed in Chapter 10. In order to objectively assess the performance of an image fusion method, a number of evaluation metrics, either objective or subjective, have been proposed. Studies on image fusion lack information that explicitly defines the applicability and feasibility of a specific fusion algorithm for a given application. Usually, a subjective evaluation is carried out to validate an objective assessment. However, identifying a reliable subjective score needs extensive experiments, which is expensive and cannot cover all possible conditions of interest. Typically, a robust performance model is required to account for the critical image fusion parameters and better assess the trend of image fusion performance quality.