Theoretical Prediction of Image Quality

Abstract

J. L. Harris has indicated that a general result from statistical decision theory can be used to derive a theoretical expression for predicting resolving power. We show that by appropriate choice of tasks, it is possible to obtain, from this same result, other measures which have previously been employed as image-quality criteria. Our results indicate that decision theory provides a common basis for deriving quality criteria for diverse tasks. Prediction of image quality for object identification is also discussed.