September 2018
Volume 18, Issue 10
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2018
Human binocular disparity estimation with natural stereo-images
Author Affiliations
  • David White
    Neuroscience Graduate Group, Biomedical Graduate Studies, University of Pennsylvania
  • Johannes Burge
    Neuroscience Graduate Group, Biomedical Graduate Studies, University of PennsylvaniaDepartment of Psychology, University of Pennsylvania
Journal of Vision September 2018, Vol.18, 993. doi:10.1167/18.10.993
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      David White, Johannes Burge; Human binocular disparity estimation with natural stereo-images. Journal of Vision 2018;18(10):993. doi: 10.1167/18.10.993.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Binocular disparity is a primary cue for depth perception. How well do humans estimate binocular disparity under natural conditions? What sources of uncertainty limit performance? What is the relative importance of each source? To answer these questions, we first sampled thousands of stereo-image patches from a recently collected stereo-database of natural scenes. Then, we used these stimuli to measure disparity discrimination performance in three human observers in a 2IFC experiment (1deg, 250ms). Each interval of every trial contained a unique contrast-fixed natural stimulus; no stimulus was presented twice. The task was to select the interval with the stimulus that was stereoscopically further away. For each of five standard disparities across an ecologically valid range (-11.25 to -3.75arcmin), we measured a nine-level psychometric function (100trl/lvl). Then, we fit each function with a cumulative Gaussian, and computed the discrimination threshold (d-prime = 1.0). Consistent with classic results, discrimination thresholds increased exponentially with disparity for each observer, confirming that the exponential law of disparity discrimination holds with natural signals. Next, each human observer repeated the experiment with the same stimuli on each trial but in a random order. This double-pass design allowed us to estimate the correlation in the observers' decision variables across the two passes of the experiment. Decision variable correlation must be stimulus-driven, because the stimuli contained the only source of uncertainty that was correlated across both passes. From the correlation we determined the ratio of uncertainty contributed by natural stimulus variability and decision noise. This ratio was approximately 1.0 for all conditions, indicating that natural stimulus variability limits performance at least as much as internal noise. Thus, natural signals have a consistent, important, and quantifiable impact on human disparity discrimination performance. Future work will model and predict the impact of natural stimulus variability on human performance this task.

Meeting abstract presented at VSS 2018

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×