June 2006
Volume 6, Issue 6
Vision Sciences Society Annual Meeting Abstract  |   June 2006
Image statistics for surface reflectance estimation
Author Affiliations
  • Lavanya Sharan
    M. I. T. Department of Electrical Engineering and Computer Science
  • Yuanzhen Li
    M. I. T. Department of Brain and Cognitive Sciences
  • Edward H. Adelson
    M. I. T. Department of Brain and Cognitive Sciences
Journal of Vision June 2006, Vol.6, 101. doi:https://doi.org/10.1167/6.6.101
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Lavanya Sharan, Yuanzhen Li, Edward H. Adelson; Image statistics for surface reflectance estimation. Journal of Vision 2006;6(6):101. https://doi.org/10.1167/6.6.101.

      Download citation file:

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

  • Supplements

Recent work has shown that certain image statistics are diagnostic of the reflectance properties of complex real-world surfaces. We have previously demonstrated that moments and percentile statistics of the luminance histogram of an image are indicative of the diffuse surface reflectance. The same statistics measured on filtered images (center-surround and oriented filters) were also shown to be useful. We have now conducted further psychophysical studies to examine the role of these statistics in reflectance perception. Subjects viewed photographs of opaque, rough real-world surfaces (e.g. crumpled paper, clay) on an LCD monitor against a middle gray background. They rated the diffuse reflectance of each surface with reference to a standard Munsell scale. We find that subjects can, to some extent, estimate the reflectance of an isolated surface in the absence of mean luminance information, contrary to the Gelb effect. Also, the perceived reflectance of complex surfaces is somewhat robust to changes in mean luminance of the image or the surround, a tendency we term as self-anchoring. Black, shiny surfaces tend to ‘self-anchor’ better than others. A learning algorithm that employs informative image statistics performs very similarly to our subjects at reflectance estimation tasks. Moreover, manipulating such statistics of an image of a real-world surface strongly affects the perceived reflectance. Therefore, taken together, our results indicate that the image statistics capture perceptually relevant information.

Sharan, L. Li, Y. Adelson, E. H. (2006). Image statistics for surface reflectance estimation [Abstract]. Journal of Vision, 6(6):101, 101a, http://journalofvision.org/6/6/101/, doi:10.1167/6.6.101. [CrossRef]
 NTT Communication Science Lab, NSF

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.