Keynote Speakers

Craig Gotsman

Craig Gotsman is a professor at the Center for Graphics and Geometric Computing (CGGC), Computer Science Department at Technion. He is also Hewlett-Packard Chair in Computer Engineering. His research is in the area of computer graphics.

His homepage is: http://www.cs.technion.ac.il/~gotsman

Geometry Processing for Computer Graphics

The field of geometry processing has developed over the past decade into a mature sub-field, alongside rendering, of computer graphics and geometric modeling.

The speaker will survey a number of key techniques in geometry processing and their application to computer graphics.

Alexei Efros

Alexei (Alyosha) Efros is an assistant professor at the Robotics Institute and the Computer Science Department at Carnegie Mellon University. His research is in the area of computer vision and computer graphics, especially at the intersection of the two. He is particularly interested in using data-driven techniques to tackle problems which are very hard to model parametrically but where large quantities of data are readily available. Alyosha received his Ph.D. in 2003 from the University of California, Berkeley and spent the following year as a fine fellow at Oxford, England. Alyosha is a recipient of the NSF CAREER award (2006), the Sloan Fellowship (2008), and the Guggenheim Fellowship (2008).

His homepage is: http://www.cs.cmu.edu/~efros/

Using Data to "Brute Force" Hard Problems in Computational Photography

(abstract pending)

Karol Myszkowski

Karol Myszkowski is a professor in Max-Planck Institute for Informatics, Germany. His research is in the area of computer graphics. He is particularly interested in high dynamic range imaging, realistic image synthesis and global illumination, perception issues in computer graphics applications.

His homepage is: http://www.mpi-inf.mpg.de/~karol/

Perception Issues in Rendering and High Dynamic Range Imaging

The knowledge of human visual system (HVS) enables more efficient image rendering by focusing the computation on visible scene details and by overcoming physical constraints of display devices.In this talk we present a number of successful examples of embedding HVS models into real-time image rendering and display pipelines. In particular, we discuss the problem of improving the appearance of highlights and light sources by boosting their apparent brightness using our temporal glare technique. Also, we show how to overcome physical contrast limitations of display device by using our 3D unsharp masking technique to boost the apparent contrast. On the image display side, we briefly overview our techniques for tone mapping of High Dynamic Range (HDR) images. Finally, we present our solutions for image quality evaluation including a difficult case of image pairs with substantially different dynamic range.

Vladimir Kolmogorov

Vladimir Kolmogorov received an MS degree from the Moscow Institute of Physics and Technology in Applied Mathematics and Physics and a PhD degree in Computer Science from Cornell University. After spending two years as an Associate Researcher at Microsoft Research, Cambridge, he joined University College London as a Lecturer. Vladimir's research is in the area of computer vision. More specifically, he focuses on graphical models, optimization algorithms for Markov Random Fields, and their applications to image segmentation, stereo and other vision problems.His papers received best paper award at ECCV 2002, best paper honourable mention award at CVPR 2005 and an honorable mention, outstanding student paper award (to M. Pawan Kumar) at NIPS 2007. He holds the Royal Academy of Engineering/EPSRC Research Fellowship.

His homepage is: http://www.cs.ucl.ac.uk/staff/V.Kolmogorov/

A Global Perspective on MAP Inference for Low-Level Vision [link]

In recent years the Markov Random Field (MRF) has become the de facto probabilistic model for low-level vision applications. However, in a maximum a posteriori (MAP) framework, MRFs inherently encourage delta function marginal statistics. By contrast, many low-level vision problems have heavy tailed marginal statistics, making the MAP-MRF framework unsuitable. I will describe a more general Marginal Probability Field (MPF), which can be used to encourage arbitrary marginal statistics.

Computing MAP solutions in an MPF is a challenging NP-hard optimization problem, which received relatively little attention in the literature. I will present several new results based around dual-decomposition and a modified min-cost flow algorithm. I will demonstrate the benefits of MPFs over MRFs on several applications, including image denoising and texture synthesis.

Joint work with Olly Woodford and Carsten Rother.