Abstract
Abstract: :
Purpose: We verify that the effects of eye movements on visual processing are common across different viewers, in particular, the temporal correlation for a given scene is the same for different viewers. Methods: We record gaze positions during free-viewing video segments of natural activities. We use the recorded eye positions to derive the gaze-centered time-varying images from the original video. We analyze the statistical properties of the resulting images, particularly, the temporal correlations. Results: Independent of viewers and scenes, there are significant temporal correlations during fixations and smooth pursuits, but the macro saccades effectively remove the temporal correlation between two signals before and after a saccade. However, the magnitudes of temporal correlations during fixations and smooth pursuits depend on the visual scenes. Most interestingly, for different viewers, the temporal correlation remains the same for any given scene, although different viewers have different gaze positions and different eye movment timings during their natural-viewing of the same scene. Conclusion: Visual input statistics during free-viewing of natural images are independent of viewers but dependent on scenes. This enables a common mechanism across viewers to improve the efficiency of visual information coding through temporal decorrelation of the input signal (Dong 2002, ARVO), and such mechanism can be dynamically adapted to different scenes to ensure high coding efficiency all the time.
Keywords: 409 eye movements: saccades and pursuits • 584 scene perception • 596 temporal vision