A Probabilistic Hypothesis for the Prediction of Visual Fixations (original) (raw)
Related papers
Visual Correlates of Fixation Selection: A Look at the Spatial Frequency Domain
2007 IEEE International Conference on Image Processing, 2007
A representation for observing local image content is proposed for the purpose of considering the distinguishing characteristics of visual content that tends to draw a human observers gaze. Within this representation, the spectral profile distinguishing fixated from non-fixated locations is considered. Finally, the possibility of designing saliency operators based on the proposed local magnitude spectrum representation is explored, revealing a promising domain for predicting human gaze patterns.
Measures and Limits of Models of Fixation Selection
PLoS ONE, 2011
Models of fixation selection are a central tool in the quest to understand how the human mind selects relevant information. Using this tool in the evaluation of competing claims often requires comparing different models' relative performance in predicting eye movements. However, studies use a wide variety of performance measures with markedly different properties, which makes a comparison difficult. We make three main contributions to this line of research: First we argue for a set of desirable properties, review commonly used measures, and conclude that no single measure unites all desirable properties. However the area under the ROC curve (a classification measure) and the KL-divergence (a distance measure of probability distributions) combine many desirable properties and allow a meaningful comparison of critical model performance. We give an analytical proof of the linearity of the ROC measure with respect to averaging over subjects and demonstrate an appropriate correction of entropy-based measures like KL-divergence for small sample sizes in the context of eye-tracking data. Second, we provide a lower bound and an upper bound of these measures, based on imageindependent properties of fixation data and between subject consistency respectively. Based on these bounds it is possible to give a reference frame to judge the predictive power of a model of fixation selection . We provide open-source python code to compute the reference frame. Third, we show that the upper, between subject consistency bound holds only for models that predict averages of subject populations. Departing from this we show that incorporating subject-specific viewing behavior can generate predictions which surpass that upper bound. Taken together, these findings lay out the required information that allow a well-founded judgment of the quality of any model of fixation selection and should therefore be reported when a new model is introduced.
Journal of Vision, 2013
Einhäuser, Spain, and Perona (2008) explored an alternative hypothesis to saliency maps (i.e., spatial image outliers) and claimed that ''objects predict fixations better than early saliency.'' To test their hypothesis, they measured eye movements of human observers while they inspected 93 photographs of common natural scenes (Uncommon Places dataset by Shore, Tillman, & Schmidt-Wulen 2004; Supplement Figure S4). Subjects were asked to observe an image and, immediately afterwards, to name objects they saw (remembered). Einhäuser et al. showed that a map made of manually drawn object regions, each object weighted by its recall frequency, predicts fixations in individual images better than early saliency. Due to important implications of this hypothesis, we investigate it further. The core of our analysis is explained here. Please refer to the Supplement for details.
Effects of spatial configuration and number of fixations on Kanizsa triangle detection
Investigative ophthalmology & visual science, 1997
Illusory figures, created by the visual system between visualizing real objects, are probably caused by processes designed to segregate objects from background. Support ratio--that is, the ratio between the physically specified and total triangle side length--has been suggested to be the main spatial determinant for suprathreshold perception of a Kanizsa-type illusion. To test this scale invariance hypothesis at threshold, illusory figure perception was studied by determining the effects of inducer size and distance at various exposure durations and fixation strategies on the frequency of seeing (FoS) an illusory Kanizsa triangle. The effect of various support ratios was studied in the first experiment by varying the intercenter distance between constant-size inducers viewed at various distances. In the second experiment, the effects of various exposure durations and fixation strategies were investigated; and the third experiment repeated the second one, with backward masking to con...
Statistical modeling of the influence of a visual distractor on the following eye-fixations
We examined the influence of a visual distractor appearing during a fixation on the following fixations during natural exploration. It is known that new objects, congruent or incongruent with the scene, appearing during a fixation are fixated more than chance [Brockmole, J. R., & Henderson, J. M. (2008). Prioritizing new objects for eye fixation in real-world scenes: Effects of object-scene consistency. Vis.Cog., 16(2-3), 375-390]. In this study, we replicated this result using a Gabor patch for th e appearing object, called a distractor because it was artificial and non-related to scenes. Besides, we wanted to quantify its influence on the xploration. A statistical model of the fixation density function was designed to analyze how the exploration was disrupted from and after the onset of the distractor. The model was composed of a linear weighted combination of different maps modeling three independent factors influencing gaze positions. We wondered whether fixation locations obser...
The role of fixational eye movements in visual perception
2004
Abstract Our eyes continually move even while we fix our gaze on an object. Although these fixational eye movements have a magnitude that should make them visible to us, we are unaware of them. If fixational eye movements are counteracted, our visual perception fades completely as a result of neural adaptation. So, our visual system has a built-in paradox—we must fix our gaze to inspect the minute details of our world, but if we were to fixate perfectly, the entire world would fade from view.
Influence of visual perception on spatial coding of saccadic eye movements and fixation
Neuro-Ophthalmology, 2004
The Müller-Lyer illusion produces differences in visual length perception when lines of equal length are presented as doublearrows with inward-pointing and outward-pointing arrowheads, respectively. The aim of this study was to investigate the influence of length perception on the amplitude of saccadic eye movements and on postsaccadic fixation positions. Video recordings of eye movements were obtained from 35 healthy volunteers (age: 20-30 years) while they scanned the endpoints of Müller-Lyer targets. The targets were presented in horizontal and vertical orientations with an arrow length of 15 •. Fixation periods between the saccades were adjusted at 1 s and 4 s. The final saccadic amplitudes between the endpoints of Müller-Lyer figures reflect the visually perceived length differences: lines with outward-pointing arrowheads were scanned with smaller saccades than lines with inwardpointing arrowheads. This bias in final saccadic amplitude was equal in both the horizontal and vertical target orientations as well as for 1-s and 4-s fixation periods. Postsaccadic drift after initial saccadic movements reduced the effect of the Müller-Lyer illusion depending on arrowhead and target orientation: outward-pointing horizontal arrows caused hypometric pulses, while inward-pointing vertical arrows caused hypermetric pulses. The results of this study demonstrate that spatial coding of both saccadic eye movements and fixation is strongly influenced by visual perception. While the influence of the Müller-Lyer illusion on final saccadic amplitudes did not depend on target orientation or fixation period, an additional pulse-step mismatch appeared in the horizontal (hypometria) and vertical (hypermetria) direction.