August 2012
Volume 12, Issue 9
Vision Sciences Society Annual Meeting Abstract  |   August 2012
Qualitative shape from shading, specular highlights, and mirror reflections
Author Affiliations
  • Michael Langer
    School of Computer Science, McGill University
  • Arthur Faisman
    School of Computer Science, McGill University
Journal of Vision August 2012, Vol.12, 232. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Michael Langer, Arthur Faisman; Qualitative shape from shading, specular highlights, and mirror reflections. Journal of Vision 2012;12(9):232.

      Download citation file:

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

  • Supplements

Surface appearance cues such as shading and specularities provide a rich source of information about 3D shape. Here we present an experiment that compared qualitative shape perception from such cues. We used OpenGL to render smooth bumpy terrain surfaces under perspective projection and under several rendering models, including the Phong model with a point light source and a mirror reflection model with a complex orientation-free environment. Each rendered terrain surface had a floor slant of 30 degrees consistent with a view-from-above prior, and each was presented either statically or rotating. The task was to judge whether a marked point on each surface was on a hill or in a valley. Percent correct scores (12 observers) were highest in the 'Lambertian only' condition, lower in the 'mirror only' and 'broad highlight only' conditions, and lowest in the 'narrow highlight only' condition. No effect was found for rotating vs. static presentation. The results are somewhat inconsistent with (Norman et al, Psych. Sci. 2004) who found that highlights improve performance in a shape discrimination task. The results are more consistent with (Fleming et al, J. Vision, 2004) who found that specularities can provide shape information if they produce a dense set of image gradients that covary with the second derivatives of depth. This requirement was met by all the rendering models above, except for 'narrow highlights only'.

Meeting abstract presented at VSS 2012


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.