🚀 go-pugleaf

	       U N I V E R S I T Y  o f  C H I C A G O
			   ---------------
		  A I   S E M I N A R   S E R I E S



DATE:   MONDAY, NOVEMBER 23
 
TIME:   2:30 PM
 
PLACE:  RYERSON 277
 
GUEST SPEAKER:  BRIAN FUNT from Simon Fraser University
 
TALK TITLE:     COMPUTATIONAL COLOR VISION
 
ABSTRACT:
 
For a long time many vision researchers seemed to feel that working
with black-and-white images removed color issues from consideration.
Black and white `seemed' simpler-- less data, less complexity.
However, ignoring color only represses many important problems since
any black-and-white picture is necessarily an image of a colorful
world and the colors in the world affect the shades of grey in the
image.  Color's effects are not limited to simple changes in albedo,
but include the less obvious color-dependent changes due to
interreflections between surfaces as well as specular reflections.  As
a result, some computer vision problems drastically simplify given the
extra information provided by color.
 
In this talk I will address two questions:
 
(i) How does color facilitate other aspects of visual perception?
 
(ii) How do we obtain color descriptors that do not depend on the
qualities of the incident illumination? 
 
In answer to these questions, I will try and present an intuitive
summary of several of my recent research results which show:
 
--how color simplifies shape from shading;
 
--how color enables the removal of confusing interreflections from
images;
 
--how color facilitates fast recognition of objects; 
 
--how color sensors can be spectrally sharpened to improve the
performance of many existing color constancy methods (eg., retinex,
CRULE, von Kries adaptation);
 
--how retinex analysis of color can be generalized to varying
resolution sensors (like the eye's) by exploiting their conformal
mapping properties.
 
                *** Tea in Ryerson 255 after the talk ***
 






[ The University of Chicago AI Laboratory sponsors a series of seminar
speakers each academic quarter. These seminars are generally of
interest to those outside the department and are open to the greater
Chicago AI community.  The department is located in Ryerson Hall at
1100 E.  58th street on the main University of Chicago quadrangle.
Seminars are usually held in Ryerson 276.]


-- 
Kristian T. Simsarian			University of Chicago AI Laboratory
email: kristian@cs.uchicago.edu		CS Dept. Chicago, IL 60637