Perception of Material Properties in 3D Scenes

Workshop to be held October 17-19, 2008
Institute for Research in Cognitive Science
University of Pennsylvania
Philadelphia, PA

Organizers

David Brainard, Larry Maloney, Anya Hurlbert

Sponsors

Institute for Research in Cognitive Science
University of Pennsylvania School of Arts and Sciences
Newcastle University
Cambridge Research Systems



Overview

For over a century, researchers have studied the perception of color and lightness for simple stimulus configurations, often equivalent to flat matte surfaces rendered under diffuse illuminations.  This foundational work has led to models of how the retinal image is encoded, and how the perceptual representations of color and lightness are related to scene properties such as object surface reflectance.

The objects we look at in daily viewing, on the other hand, are rarely flat, matte, or diffusely illuminated.  This observation has led to considerable recent interest about how we perceive object reflectance in more complex, three-dimensional scenes.  For example, the mug shown to the right is readily perceived as blue.  Yet there is large variation in reflected light across the surface of the mug, as shown by the three squares inset at the top of the figure.  Each of these squares was taken from a single point on the mug, raising the question of how the visual system integrates the light reflected from multiple locations to produce the perception that the surface of the mug is homogeneous, smooth, shiny and blue.

The questions that arise from considering object surface properties in 3D scenes sit at the intersection of the fields of visual perception, computer vision, and computer graphics.  They include a) what is the effect of object material (e.g. wood, plastic, metal) on color and lightness perception (visual perception), b) how are object reflectance properties best measured and parameterized (computer graphics and computer vision), c) what are good models of image formation for complex scenes (computer graphics and computer vision), d) how do we perceive what materials an object is made of (visual perception), e) how do we integrate chromatic information across the image of an object to arrive at unified percept of the object’s color (visual perception), f) what information that might allow separation of object and illuminant properties is available in the image (computer vision), and g) is this information used by human vision (visual perception)?  Work on these questions has already led to new consideration of the interactions between object shape and orientation, object material, and the distribution of illumination in a scene (“the light field”), and to led to thinking about color, lightness, gloss, and visual roughness as a set of perceptual attributes that together inform us about object properties, rather than dimensions that should be studied separately.

The objective of this workshop is to bring together an interdisciplinary group researchers working at the forefront of this emerging, exciting domain.  As described below, we believe that at this juncture we can leverage the workshop to help define and shape a research agenda for the field that will enable much more rapid and collaborative progress than if the relevant individual investigators proceed without close communication.

Invited Speakers

The confirmed invited speakers are Bart Anderson, Marina Bloj, David Brainard, Mike Chantler, Hany Farid, Roland Fleming, David Foster, Karl Gegenfurtner, Alan Gilchrist, Anya Hurlbert, Larry Maloney, Shin'ya Nishida, Sylvia Pont, and Todd Zickler.

Registration

Note: The registration site will close Thursday October 9. We may be able to accept cash or check registrations at the workshop on Friday, October 17. Please email Laurel Sweeney (laurels@seas.upenn.edu) in advance if you plan to arrive without having registered, so that we can confirm that this option will be available. Penn students and faculty may attend the talks without registering (or paying), but should still let Laural know so that we can make sure to have enough coffee.

Registration for workshop attendees is $80.00. This registration fee includes access to all activities and talks, as well as a boxed lunch on Friday. The Saturday dinner is now full, we cannot take any additional guests at the dinner.

To register for the workshop, please go to https://www.seattletech.com/registrations/index.php?897-10013-i-t. The site will also allow you to submit a poster for presentation.

If you have trouble with the registration site, please email Laurel Sweeney with your questions.

You will need to make your own arrangements for accommodation. The registration site provides information about nearby hotels.