Can image quality features predict visual change blindness?
Steven Le Moan · 2017
This study concerns the prediction of a striking perceptual failure known as change blindness, which affects our ability to notice certain types of differences between visual stimuli (e.g. change in the colour or position of an object). We draw a parallel with the subjective image fidelity assessment task, which also consists of determining differences between images. Here, we look at whether the features extracted by recent suprathreshold indices can effectively predict the difficulty of a change detection task. Both linear and nonlinear feature pooling approaches are tested and validated on two databases. Results indicate that image quality features show promise when it comes to predicting detection times.