How the task of evaluating image quality influences viewing behavior

Abstract

Image quality scores collected in subjective experiments are widely used in image quality research, particularly in the design of objective quality assessment algorithms. It is therefore of vital importance to make sure that the collected scores reflect viewers' opinions in real-life situations. However, just by giving the viewers the task of assessing quality, there is a risk that their behavior has changed from what it would be in a natural viewing condition. We here investigate this difference in behavior by examining the gaze response in both conditions, i.e. free looking and scoring quality, with the help of eye-tracking equipment. Even though the observed behavior shows similarities between the two conditions, there are also significant differences which should be taken in consideration in future image quality research.