Assessment of Lossy Images with Visual Detection and Saccadic Eye Movements

This study investigated a novel method of evaluating visually lossy images based on saccadic eye movements. In each trial, participants with normal vision were asked to indicate any visible changes in the image while their gaze positions were being monitored. The original image was replaced with compressed or blurred versions of the same image 150ms after the onset of each eye fixation, and the parameters of the immediately following saccades were assessed. These outcomes were compared to those with a constant rate of image switch (5 Hz flicker) and without switch (control). The resultant rate of visual detection was higher with blurred and constant flickering images than compressed images. Changes in fixation duration, saccade amplitude and peak saccade velocity suggest disrupted visuomotor processes associated with image switch. These disruptions occurred with or without conscious detection. The use of flickering images likely overestimates the probability of detecting lossy image. Analysis of eye movements might be more useful in discerning lossy images in situations where various form factors and visuomotor contexts need to be evaluated.