Spatial Color Perception and Image Processing
Instructor: Edoardo Provenzi, Université de Bordeaux
Level: Intermediate
Duration: 2 hours plus 15 minute break. After the class, adjourn to Zoom to join the instructor and other students in a discussion of the class.
Course Time:
    New York: Thursday 12 November, 10:00-12:15
    Paris: Thursday 12 November, 16:00-18:15
    Tokyo: Friday 13 November 00:00-02:15

Benefits:
Attendees will be able to:

• Understand the important perceptual phenomenon of visual induction.
• Learn a simple mathematical model for color induction and how to implement it through psychophysical experiments.
• Be introduced to the very powerful concept of variational model, universally used in imaging.

Intended Audience: the course is intended for a very broad audience: master and PhD students, post-doc researchers, scientists or engineers working in the following disciplines: color science, colorimetry, image processing (especially spatially-varying color enhancement and tone mapping of HDR images), psychophysics or cognitive, and perceptual psychology. More generally, anyone who wishes to get acquainted with some very modern theories of color vision, based on the rediscovery of some profound and beautiful works of our ancestors. This modern vision puts in evidence the need of performing new and fascinating psychophysical experiments which have the potential to lead to the future era of colorimetry. A general knowledge of calculus and linear algebra is required to fully grasp the concepts that are discussed. A basic knowledge of psychophysics is not necessary but it would be helpful for the sake of a better interaction.

Course Description:
Spatial induction, i.e. the modification of color perception of a point caused by a non-uniform surround, plays an important role in image processing. To quote just a few applications, it is used as a main ingredient to devise color enhancement algorithms and tone mapping operators in the high dynamic range imaging field. The first hour provides a detailed description of a psychophysical technique for measuring the relative strength of spatial induction focusing, at first, on a very simple geometrical configuration and then extending it to more complex and realistic ones.
During the following hour, we analyze a variational model that formalize and extends the previous description to a more general framework that accounts also for cognitive phenomena."

Edoardo Provenzi obtained a master in theoretical physics at the University of Milan (2000) and a PhD in applied mathematics (2004) at the University of Genoa. Since 2004, he has been studying color vision and image processing. He is currently working in these research fields: color science, geometry of color space, statistics of natural images, variational techniques in imaging. He worked as associate professor from 2014 to 2017 at the University Paris Descartes. He is Full Professor of Applied Mathematics in the Image Processing Group at the University of Bordeaux since 2017 where he is in charge of the Master program on Mathematics for signals and images. He is the author of the book: Computational Color Science: Variational Retinex-like Methods, (Wiley & Sons Eds. 2017).

For office use only: