Model-based quantification of image quality

In 1982, Park and Schowengerdt published an end-to-end analysis of a digital imaging system quantifying three principal degradation components: (1) image blur - blurring caused by the acquisition system, (2) aliasing - caused by insufficient sampling, and (3) reconstruction blur - blurring caused by the imperfect interpolative reconstruction. This analysis, which measures degradation as the square of the radiometric error, includes the sample-scene phase as an explicit random parameter and characterizes the image degradation caused by imperfect acquisition and reconstruction together with the effects of undersampling and random sample-scene phases. In a recent paper Mitchell and Netravelli displayed the visual effects of the above mentioned degradations and presented subjective analysis about their relative importance in determining image quality. The primary aim of the research is to use the analysis of Park and Schowengerdt to correlate their mathematical criteria for measuring image degradations with subjective visual criteria. Insight gained from this research can be exploited in the end-to-end design of optical systems, so that system parameters (transfer functions of the acquisition and display systems) can be designed relative to each other, to obtain the best possible results using quantitative measurements.