Image Sciences, Computer Sciences and Remote Sensing Laboratory

Next seminars

Thursday, March 18 2010 - 10:30 am


Auto-organisation appliquée à la localisation pour réseaux de capteurs Lieu : Salle C427 Conférencier(s) : Karel Heurtefeux (INSA Lyon, Laboratoire CITI)

Past seminars :

Friday, March 05 2010 - 2:00 pm

Mathematics and Global Illumination Lieu : Amphi A508 (ENSPS) Conférencier(s) : Massimo Picardello , Professor of Mathematics, University of Rome "Tor Vergata", Italy Résumé : This is a simple, almost naif approach to the mathematics of global illumination in Computer Graphics. The speaker is by no way an expert: this presentation is only aimed to sketch how this material is explained to students of mathematics in Roma ``Tor Vergata'' with an interest for computer graphics. Classical Computer Graphics brought to two leading iterative methods for realistic rendering: Recursive ray Tracing (where bounces of light are traced ray after ray following their reflections), and Radiosity (a finite element method for global light interchange). The former gives approximate illumination with added precise highlights, the latter fine light gradation fore purely diffuse environments without highlights or directional effects. Modern Computer Graphics rediscovers these two methods in one unified photorealistic approach, Global Illumination, based on recursive high-dimensional integral equations solved by tracing rays with suitably chosen probability distribution and computing the integrals with stochastic sampling.

Mathematics and Global Illumination

Lieu : Amphi A508 (ENSPS)

Conférencier(s) : Massimo Picardello , Professor of Mathematics, University of Rome "Tor Vergata", Italy

Résumé : This is a simple, almost naif approach to the mathematics of global illumination in Computer Graphics. The speaker is by no way an expert: this presentation is only aimed to sketch how this material is explained to students of mathematics in Roma ``Tor Vergata'' with an interest for computer graphics. Classical Computer Graphics brought to two leading iterative methods for realistic rendering: Recursive ray Tracing (where bounces of light are traced ray after ray following their reflections), and Radiosity (a finite element method for global light interchange). The former gives approximate illumination with added precise highlights, the latter fine light gradation fore purely diffuse environments without highlights or directional effects. Modern Computer Graphics rediscovers these two methods in one unified photorealistic approach, Global Illumination, based on recursive high-dimensional integral equations solved by tracing rays with suitably chosen probability distribution and computing the integrals with stochastic sampling.

Friday, December 18 2009 - 10:00 am

Research at Philips Hamburg on organ segmentation and registration for radiotherapy Lieu : Amphi A 301 (ENSPS) Conférencier(s) : Torbjorn VIK, Philips (Hambourg). Résumé : In this presentation I will give an overview of radiotherapy related image processing at Philips Research in Hamburg. After a short introduction on Philips Research, I will get into the background and the clinical motivation for automatic image segmentation and registration in this domain. This is followed by a technical part where I first present a deformable surface algorithm for organ segmentation, followed by a presentation of a fast registration framework for model initialization. These algorithms are used for automatic segmentation of structures in the clinically relevant sites head and neck, thorax and pelvis. Finally, I would like to go into some of the more academic issues that we are studying which are 4D-CT lung registration and the propagation of segmentation uncertainty to dose volume histograms.

Research at Philips Hamburg on organ segmentation and registration for radiotherapy

Lieu : Amphi A 301 (ENSPS)

Conférencier(s) : Torbjorn VIK, Philips (Hambourg).

Résumé : In this presentation I will give an overview of radiotherapy related image processing at Philips Research in Hamburg. After a short introduction on Philips Research, I will get into the background and the clinical motivation for automatic image segmentation and registration in this domain. This is followed by a technical part where I first present a deformable surface algorithm for organ segmentation, followed by a presentation of a fast registration framework for model initialization. These algorithms are used for automatic segmentation of structures in the clinically relevant sites head and neck, thorax and pelvis. Finally, I would like to go into some of the more academic issues that we are studying which are 4D-CT lung registration and the propagation of segmentation uncertainty to dose volume histograms.

Archives:

Academic Year 2008/2009
Academic Year 2007/2008
Academic Year 2006/2007
Academic Year 2005/2006
Academic Year 2004/2005