Everyone knows what is interesting: salient locations which should be fixated - PubMed Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2009 Oct 27;9(11):25.1-22.
doi: 10.1167/9.11.25.

Everyone knows what is interesting: salient locations which should be fixated

Christopher Michael Masciocchi  1 , Stefan Mihalas, Derrick Parkhurst, Ernst Niebur
Affiliations

Everyone knows what is interesting: salient locations which should be fixated

Christopher Michael Masciocchi et al. J Vis. .

Abstract

Most natural scenes are too complex to be perceived instantaneously in their entirety. Observers therefore have to select parts of them and process these parts sequentially. We study how this selection and prioritization process is performed by humans at two different levels. One is the overt attention mechanism of saccadic eye movements in a free-viewing paradigm. The second is a conscious decision process in which we asked observers which points in a scene they considered the most interesting. We find in a very large participant population (more than one thousand) that observers largely agree on which points they consider interesting. Their selections are also correlated with the eye movement pattern of different subjects. Both are correlated with predictions of a purely bottom-up saliency map model. Thus, bottom-up saliency influences cognitive processes as far removed from the sensory periphery as in the conscious choice of what an observer considers interesting.

PubMed Disclaimer

Figures

Figure 1
Figure 1
One example of each image from the four categories: A) buildings, B) fractals, C) home interiors, and D) landscapes.
Figure 2
Figure 2
Mean reaction time to make interest point selections. Error bars represented plus and minus one standard error.
Figure 3
Figure 3
Two example images from each image category demonstrating the clustering of interest points. First selections are shown as red dots, selections two to five as blue dots.
Figure 4
Figure 4
Grand average (over all images) of interest point selections (left) and fixations (right). The strong central bias is discussed in the Summary of results of Experiment 2.
Figure 5
Figure 5
Clustering of interest points determined by the percent of interest point selections a given distance (pixels) away from each other interest point for the four image categories. Error bars represent plus and minus one standard error.
Figure 6
Figure 6
Clustering of interest points determined by the percent of interest points that fell within an interest cluster for the four image categories. Error bars represent plus and minus one standard error.
Figure 7
Figure 7
Creation of interest maps. (A) Original image with interest points plotted on top. Color coding as in Figure 3. (B) Interest selections with Gaussian intensity “blobs” centered on each interest point and superposed. Values have been normalized such that the sum of all values equals unity, and the overall mean has been subtracted.
Figure 8
Figure 8
(A) Mean saliency values at different interest point locations for the four image types and the actual and chance sampling distributions. Error bars represent plus and minus one standard error. Note that the distance between the actual and chance sampling distributions represents the chance-adjusted saliency value. (B) Cross-correlation between interest maps and saliency maps. Values from the random cross-correlation distribution are sorted from weakest to strongest and from negative to positive correlations. The mean value of the actual distribution is plotted in green.
Figure 9
Figure 9
(A) Mean saliency values at different fixation locations for the four image types and the actual and chance sampling distributions. Error bars represent plus and minus one standard error. Note that the distance between the actual and chance sampling distributions represents the chance-adjusted saliency value. (B) Cross-correlation between fixation maps and saliency maps. Values from the random cross-correlation distribution are sorted from weakest to strongest and from negative to positive correlations. The mean value of the actual distribution is plotted in green.
Figure 10
Figure 10
(A) Mean interest values at different fixation locations for the four image types and the actual and chance sampling distributions. Error bars represent plus and minus one standard error. Note that the distance between the actual and chance sampling distributions represents the chance-adjusted interest value. (B) Cross-correlation between fixation maps and interest maps. Values from the random cross-correlation distribution are sorted from weakest to strongest and from negative to positive correlations. The mean value of the actual distribution is plotted in green.
Figure 11
Figure 11
Example image (A) with associated interest map (B), fixation map (C), and saliency map (D).
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Bacon WF, Egeth HE. Overriding stimulus-driven attentional capture. Perception & Psychophysics. 1994;55:485–496. [PubMed] - PubMed
    1. Cerf M, Cleary DR, Peters RJ, Einhäuser W, Koch C. Observers are consistent when rating image conspicuity. Vision Research. 2007;47:3052–3060. [PubMed] - PubMed
    1. Craft E, Schütze H, Niebur E, von der Heydt R. A neural model of figure–ground organization. Journal of Neurophysiology. 2007;97:4310–4326. [PubMed] [Article] - PubMed
    1. Driver J, Baylis G. Edge-assignment and figure–ground segmentation in short-term visual matching. Cognitive Psychology. 1996;31:248–306. [PubMed] - PubMed
    1. Elazary L, Itti L. Interesting objects are visually salient. Journal of Vision. 2008;8(3):3. 1–15, http://journalofvision.org/8/3/3/, doi:10.1167/8.3.3. [PubMed] [Article] - PubMed

Publication types