September 2018
Volume 18, Issue 10
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2018
Modeling 3D Slant Perception: Bootstrapping 3D Affine Structure to Euclidean
Author Affiliations
  • Xiaoye Wang
    Department of Psychological and Brain Science, Indiana University Bloomington, IN, USA
  • Mats Lind
    Department of Information Technology, Uppsala University, Uppsala, Sweden
  • Geoffrey Bingham
    Department of Psychological and Brain Science, Indiana University Bloomington, IN, USA
Journal of Vision September 2018, Vol.18, 129. doi:10.1167/18.10.129
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Xiaoye Wang, Mats Lind, Geoffrey Bingham; Modeling 3D Slant Perception: Bootstrapping 3D Affine Structure to Euclidean. Journal of Vision 2018;18(10):129. doi: 10.1167/18.10.129.

      Download citation file:


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

      ×
  • Supplements
Abstract

Lind et al. (2014) proposed a Bootstrapping model to account for results of using large continuous perspective change to recover the unknown affine scaling factor in the perception of 3D polyhedral shape. The model assumes 3D affine structure and bootstraps up to Euclidean. The current study extended application of this model to 3D slant perception using monocular optic flow. Because non-coplanar points (not typical in slant displays) are theoretically required for the affine structure, we tested slant judgments of strictly planar surfaces compared to surfaces with non-coplanar points. We compare simulations with empirical results from human observers to evaluate the model's effectiveness. Methods We simulated planar surfaces defined by a series of points, both in displays and in model simulations. We used perspective projections to obtain visual coordinates and relevant information associated with the surface points, to which we applied the bootstrapping model to derive predicted slants as a function of different amounts of continuous perspective change and visual noise. We compared the simulation results with judgment results. Results We found that the model failed to generate accurate predictions of slant when the surface was planar, lacking non-coplanar points. However, when non-coplanar points were introduced, the model prediction became accurate. As predicted, judgments became accurate in the face of representative levels of visual noise with ≥ 45° perspective change. Conclusion The Bootstrapping model and empirical data showed that first, optical information allows perception of 3D affine (or relief) structure, and secondly, sufficiently large perspective change enables application of affine operations described by the model to bootstrap affine structure to Euclidean structure via the required scaling constant.

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.

×