This article is more than 5 years old. Information may no longer be current.
Background luminance, contrast, spacing affect ability to focus on an object
Click Here to Manage Email Alerts
“The visibility of a target object may be affected by the specific properties of the background scene at and near the target’s location and by how uncertain the observer is from one occasion to the next about the values of the background and target properties,” according to a study from Stephen Sebastian, PhD, and fellow researchers at the University of Texas.
Visual detection thresholds increase approximately linearly along all three dimensions, and detection accuracy is unaffected by background bin and target amplitude uncertainty, according to research published in the Proceedings of the National Academy of Sciences of the United States of America.
The researchers’ experimental method utilized constrained sampling from multidimensional histograms of natural stimuli and a theoretical analysis based on signal detection theory.
Researchers organized a large collection of natural image backgrounds into multidimensional histograms with the bins corresponding to a specific luminance, contrast and similarity.
Detection thresholds were measured for a subset of the bins where natural backgrounds were randomly sampled from a bin on each trial.
In low certainty conditions, both the background bin and the amplitude of the target were fixed, and in high-certainty conditions, they varied randomly on each trial, researchers wrote.
Sebastian and colleagues wanted to determine how luminance, contrast, spatial similarity and uncertainty due to random variations can individually affect detection accuracy in natural scenes; how they combine in affecting detection accuracy; and how the factors and underlying neural mechanisms are related to the statistical properties of natural scenes.
Amplitude thresholds increase approximately linearly along each of the cardinal directions for detection in natural backgrounds, they wrote, with different slopes and intercepts for each dimension and target.
This is true both when the background region extended well beyond the target region and when the background was restricted to the target area.
The researchers found that human thresholds in natural backgrounds are accurately predicted from first principles.
It is understood that for detection in white noise as a function of noise amplitude, the effect of removing a fraction of pixels in a matched template reduces efficiency by a fixed scale factor without affecting the pattern of thresholds, according to the researchers.
They tested this hypothesis within natural images and found that it holds well.
“Under natural conditions, both the background and the amplitude of the target (if present) are generally different on every occasion,” the researchers wrote.
“What [we show] is that the detrimental effects of this uncertainty can be optimally reduced by dividing the template response by the product of background luminance, contrast and similarity,” they continued.
They concluded that the rapid and local neural gain-control mechanisms and the psychophysical laws of masking are likely the result of evolving a near-optimal solution to detection in natural backgrounds under conditions of high uncertainty. – by Abigail Sutton
Disclosures: The researchers report no relevant financial disclosures.
Click Here to Manage Email Alerts