= PDF Reprint, = BibTeX entry, = Online Abstract
P. Tseng, I. G. M. Cameron, D. P. Munoz, L. Itti, Effects of development on low-level feature processing during natural viewing of dynamic scenes, In: Proc. Vision Science Society Annual Meeting (VSS11), May 2011.
Abstract: Eye movements have been widely used to examine many aspects of brain functions, such as reflexive response, inhibitory controls, and working memory, in normal development. However, it is unclear how normal development affects eye movements of natural viewing behavior. This study specifically examined the developmental trajectory of low-level features processing while participants freely viewed videos of natural scenes. These videos are composed of short (2–4 seconds), unrelated clips. This design was to reduce top-down expectation and to magnify the difference in gaze allocation at every scene change. Gazes of 3 groups of participants (18 children, 10.7 +/- 1.8 yr; 18 young adults, 23.2 +/- 2.6 yr; 24 elderly, 70.3±7.5 yr) were tracked while they watched the videos for 20 minutes. First, we used a computational saliency model (Itti & Koch, 2001) to compute bottom-up saliency maps for each video frame. These saliency maps can be computed from a single feature (e.g. color contrast, motion contrast) or a combination of them. Next, we computed the correlation between salience and gaze of each population. To reveal the developmental trajectory of low-level features processing, classifiers were built to differentiate (1) children vs. young adults, and (2) young adults vs. elderly. In the mean time, a feature selection method was performed to identify the most discriminative features for differentiating the populations. Using this method, we found that during normal maturation (children to young adults), there was a reduction in saccade interval and an increase in correlation between gaze and texture contrast, orientated edges, and color contrast. On the other hand, during normal aging (young adults to elderly), we found an increase in saccade interval and a decrease in correlation between gaze and oriented edges. In conclusion, this study revealed for the first time the differences between age groups in low-level feature processing during natural viewing of dynamic scenes.
Themes: Human Psychophysics, Computational Modeling, Human Eye-Tracking Research
Copyright © 2000-2007 by the University of Southern California, iLab and Prof. Laurent Itti.
This page generated by bibTOhtml on Thu Jan 31 11:39:41 PST 2019