The relationship between the locations of spatial features and those of fixations made during visual examination of briefly presented images

Abstract

The locations of features such as extremes of contrast or luminance, high spatial frequency content and edge density in a set of images have been determined, and the locations of fixations made by a group of eighteen human observers who examined the images during brief (3 s) presentations were also measured. The similarity between the locations of the eye movements and those of each stimulus feature was determined by means of a least squares index IS. For averages taken over data for all observers, the similarity determined in this way is much lower than values for pairs of fixation locations made by different observers. It is concluded that the pattern of fixations made to a given image which is highly conserved between different observers, cannot be associated with any one of the local features examined by us. It is shown further that the distribution of fixation locations over the images is non-uniform, with a marked bias to central areas, whereas the image features are more uniformly distributed. Weighting the distributions of feature locations to take account of the non-uniform distribution of fixations produces much higher IS values, but the dominant contribution to these high values is the weighting function itself. Only in the case of edge density is there significant similarity between the locations of eye movements and those of the image features.

Eye movements in reading and information processing: 20 years of research.

“…Furthermore, landing position influenced fixation time: The duration of the first fixation on an object decreased and the probability of refixating an object increased as the deviation of the initial landing position from the center of the object increased. Mannen et al (1995Mannen et al ( , 1996Mannen et al ( , 1997 had viewers look at normal scenes or the scenes after they had been either high-pass or low-pass filtered. They found that viewers looked in pretty much the same locations across the three versions of the scene.…”

Visual saliency does not account for eye movements during visual search in real-world scenes

“…Edge density was defined as the proportion of edge pixels within an image patch. Edge pixels were found by filtering the scenes with a Sobel operator that responds to contours in scenes represented by steep gradients in luminance (see Mannan et al, 1995Mannan et al, , 1996. Greater edge density is associated with image patches containing a greater number of contours.…”

“…For example, high spatial frequency content and edge density have been found to be somewhat greater at fixated than non-fixated locations (Mannan, Ruddock, & Wooding, 1996, 1997b. Furthermore, local contrast (the standard deviation of intensity in a patch) is higher and two-point intensity correlation (intensity of the fixated point and nearby points) is lower for fixated scene patches than control patches (Krieger, Rentschler, Hauske, Schill, & Zetzsche, 2000;Parkhurst & Neibur, 2003;Reinagel & Zador, 1999).…”

“…Several studies have demonstrated that the image properties of fixated regions tend to differ in systematic ways from regions that are not fixated (Krieger et al, 2000;Mannan, et al, 1995Mannan, et al, , 1996Mannan, Ruddock, & Wooding, 1997a;Parkhurst & Niebur, 2003;Reinagel & Zador, 1999). Specifically, fixated scene regions tend to be lower in intensity but higher in edge density and local contrast, and are more likely to contain third-order spatial relationships such as T-junctions and curves, than non-fixated regions.…”

Initial fixation placement in face images is driven by top–down guidance

“…External factors are image immanent features, such as local image contrast and local image structure, which transiently attract eye gaze, independent of a particular task. Internal factors, such as an individual's attentional state, expectation, experience and memory, are top-down and task-dependent (Noton and Stark 1971;Mannan et al 1997;Henderson 2003). It is argued that in general, the initial saccade to an image is driven predominantly by external factors (Parkhurst et al 2002;Peters et al 2005), but can also be biased by internal factors (Henderson 2003).…”