Quality evaluation in spectral imaging – Quality factors and metrics

Abstract

Spectral imaging has many advantages over conventional three channel colour imaging, and has numerous applications in many domains. Despite many benefits, applications, and different techniques being proposed, little attention has been given to the evaluation of the quality of spectral images and of spectral imaging systems. There has been some research in the area of spectral image quality, mostly targeted at specific application domains. This paper seeks to provide a comprehensive review on existing research in the area of spectral image quality metrics. We classify existing spectral image quality metrics into categories based on how they were developed, their main features, and their intended applications. Spectral quality metrics, in general, aim to measure the quality of spectral images without considering specifically the imaging systems used to acquire the images. Having many different types of spectral imaging systems that could be used to acquire spectral images in an application, it is important to evaluate the performance/quality of these spectral imaging systems too. However, to our knowledge, not much attention has been given in this direction previously. As a first step towards this, we aim to identify different factors that influence the quality of the spectral imaging systems. In almost every stage of a spectral imaging workflow, there may be one or more factors that influence the quality of the final spectral image, and hence the imaging system used for acquiring the image. Identification of these factors, we believe, will be essential in developing a framework, for evaluating the quality of spectral imaging systems.