Visual correlates of fixation selection: effects of scale and time

Abstract

What distinguishes the locations that we fixate from those that we do not? To answer this question we recorded eye movements while observers viewed natural scenes, and recorded image characteristics centred at the locations that observers fixated. To investigate potential differences in the visual characteristics of fixated versus non-fixated locations, these images were transformed to make intensity, contrast, colour, and edge content explicit. Signal detection and information theoretic techniques were then used to compare fixated regions to those that were not. The presence of contrast and edge information was more strongly discriminatory than luminance or chromaticity. Fixated locations tended to be more distinctive in the high spatial frequencies. Extremes of low frequency luminance information were avoided. With prolonged viewing, consistency in fixation locations between observers decreased. In contrast to [Parkhurst, D. J., Law, K., & Niebur, E. (2002). Modeling the role of salience in the allocation of overt visual attention. Vision Research, 42 (1), 107-123] we found no change in the involvement of image features over time. We attribute this difference in our results to a systematic bias in their metric. We propose that saccade target selection involves an unchanging intermediate level representation of the scene but that the high-level interpretation of this representation changes over time.

Are fixations in static natural scenes a useful predictor of attention in the real world?

“…Here, we focused on the overall spatial similarity between regions inspected in the different conditions by using fixation density maps (Wooding, 2002) and the area under the Receiver Operating Characteristics (ROC) curve (t 'Hart et al, 2009;Ehringer et al, 2009;Tatler et al, 2005). This method has become a standard approach for testing computational models of fixation location, and it has a number of advantages because it makes no assumptions about differences in the underlying distributions.…”

Compressed-domain visual saliency models: a comparative study

“…2) Center Bias and Border Effects: A person recording a video will generally tend to put regions of interest near the center of the frame [73], [79]. In addition, people also have a tendency to look at the center of the image [78], presumably to maximize the coverage of the displayed image by their field of view.…”

Bottom–up attention: pulsed PCA transform and pulsed cosine transform

“…A number of recent papers (e.g., Bruce and Tsotsos 2005;Harel et al 2006;Gao et al 2007;Tatler et al 2005) used receiver operating characteristic (ROC) curve to evaluate a saliency map's ability to predict human eye fixations. Given a threshold value, a saliency map can be divided into the target region and background region.…”