T8

Applications of Information Theory to Computer Graphics

---

Date: Tuesday, September 4^th, Time: 14:00 - 17:30, Location: Lecture Room 309

---

Organizers:		Mateu Sbert (University of Girona, Spain) Miquel Feixas (University of Girona, Spain)
Speakers:		Mateu Sbert (University of Girona, Spain) Miquel Feixas (University of Girona, Spain) Jaume Rigau (University of Girona, Spain) Ivan Viola (University of Bergen, Norway) Miguel Chover (Jaume I University, Spain)

---

Abstract

We present different applications of information theory to computer graphics, based on the use of the measures of entropy, mutual information, f-divergences and generalized entropies. The application areas are hierarchical radiosity, adaptive pixel supersampling, selection of best viewpoints, object and scene exploration, mesh saliency, mesh simplification and scientific visualization. We also give some hints on information-theoretic applications to object recognition and image processing.


    

    

---

Syllabus

1. Introduction (5 min)
   
   Speaker: Mateu Sbert
   
   
2. Information Theory Basics (35 min)
   
   Speaker: Miquel Feixas
   
   
   • Information channel: entropy and mutual information
   • Important inequalities
   • Information bottleneck method


3. Refinement Criteria for Radiosity (20 min)
   
   Speaker: Jaume Rigau
   
   
   • Scene information channel
   • Refinement criteria for hierarchical radiosity
   • Mutual-information-based oracle
   • f-divergence-based oracles


4. Adaptive Refinement for Ray-tracing (15 min)
   
   Speaker: Jaume Rigau
   
   
   • Refinement criteria for ray-tracing
   • Entropy-based refinement criteria
   • f-divergence-based refinement criteria


5. Viewpoint Selection and Mesh Saliency (30 min)
   
   Speaker: Mateu Sbert
   
   
   • Viewpoint information-theoretic measures
   • Viewpoint information channel: mutual information, similarity and stability
   • Selection of best views and object exploration
   • Polygonal mutual information and information-theoretic ambient occlusion
   • Mesh saliency
   • Importance-driven viewpoint selection


6. View Selection in Scientific Visualization (30 min)
   
   Speaker: Ivan Viola
   
   
   • View Selection for Volumes and Iso-Surfaces
   • Importance-Driven Focus of Attention
   • Guided Navigation using Higher-Level Semantics


7. Viewpoint-driven Simplification (30 min)
   
   Speaker: Miguel Chover
   
   
   • Recent work on simplification
   • Information-theoretic metrics: entropy, Kullback-Leibler distance and mutual information
   • Simplification algorithms


8. Other Applications (15 min)
   
   Speaker: Miquel Feixas and Mateu Sbert
   
   
   • Image processing: registration and segmentation
   • Object recognition: shape descriptors
   • Other

---

Speakers' Background

Mateu Sbert is a professor in Computer Science at the University of Girona, Spain. He received a M.Sc. in Theoretical Physics (1977) at the university of Valencia, a M.Sc. in Mathematics (Statistics and Operations Research, 1983) at U.N.E.D. university (Madrid) and a Ph.D. in Computer Science at the U.P.C. (Universitat PolitŹcnica de Catalunya, 1997, Best PhD Award). Mateu Sbert's research interests include the application of Monte Carlo, Integral Geometry and Information Theory techniques to Radiosity, Global Illumination and Image Based Rendering. He has authored or co-authored more than one hundred papers in his areas of research, participated in several Eurographics tutorials and served as a member of program committee in Spanish and international conferences. He has participated in several European and Spanish research projects, leading the VIth European Framework Gametools project. Mateu Sbert coorganized the Dagsthul Seminars "Stochastic Methods in Rendering" and "Computational Aesthetics in Graphics, Visualization and Imaging".

Miquel Feixas is an associate professor in Computer Science at the University of Girona, Spain. He received a M.Sc. in Theoretical Physics (1979) at the UAB (Universitat Autonoma de Barcelona) and a Ph.D. in Computer Science at the UPC (Universitat PolitŹcnica de Catalunya, 2002). His research is focused on the application of Information Theory techniques to Radiosity, Global Illumination, Viewpoint Selection and Image Processing. He has co-authored several papers in his areas of research. He acted as a reviewer for conferences in the field of computer graphics and image processing. He has participated in Spanish research projects and joint actions with several European universities.

Jaume Rigau is an associate professor in Computer Science at the University of Girona, Spain. He received a M.Sc. in Computer Science (1993) at the UPC (Universitat Polit cnica de Catalunya) and a Ph.D. in Computer Science at the UPC (Universitat PolitŹcnica de Catalunya, 2006). His research is focused on the application of Information-Theory techniques to Radiosity, Global Illumination and Image Processing. He has co-authored several papers in his areas of research. He has participated in several Spanish research projects and joint actions with several European universities.

Ivan Viola is a Post Doctor research associate at University of Bergen, Norway. He was formerly associated with Vienna University of Technology, Austria, where he received M.Sc. in 2002 and Ph.D. in 2005. His research is focused on development of novel methods for automatically generating expressive visualizations of complex data. Viola co-authored several scientific works published on international conferences such as IEEE Visualization, EuroVis, and Vision Modeling and Visualization and acted as a reviewer and program committee member for conferences in the field of computer graphics and visualization. Recently he co-organized series of tutorials on Illustrative Visualization.

Miguel Chover is an associate professor at the Universitat Jaume I de Castello, Spain. His research focused on interactive computer graphics, computer games and Web3D. His current work includes level of detail modelling, simplification algorithms, rendering natural phenomena and texturing techniques. He received his MS degree in Computer Science in 1992 and a PhD in Computer Science in 1996 from the Universidad Politecnica de Valencia, Spain. He is member of Eurographics.



« Back