Lecture Series in Pattern Recognition
题 目（TITLE）：Applications of Information Theory to Global Illumination, Shape Recognition and Image Processing
讲 座 人（SPEAKER）：Prof. Mateu Sbert (University of Girona, Spain)
主 持 人 (CHAIR)：Prof. Xiaopeng Zhang
时 间 (TIME)：10:00AM, Apr 24 (Tuesday), 2012
地 点 (VENUE)：The Second Meeting Room, 13th floor
We introduce scene complexity measures and their application to radiosity. Radiosity is a viewpoint independent global illumination technique that discretizes the scene into small polygons or patches to solve a transport system of equations. The way the scene is discretized is critical for the efficiency of the result. First, we define a scene information channel, which allows us to study the interchange of information between the patches. From the study of this channel, several refinement oracles, i.e., criteria for subdividing the geometry, are obtained, aimed at maximizing the transport of information. We also present adaptive ray tracing. This technique is aimed at tracing more rays only where they are needed. Information-theoretic measures, such as Shannon entropy, Tsallis entropy, and f-divergences, will be used to define adaptive refinement criteria.
Another application of information-theoretic measures is to obtain different shape descriptors based on the complexity of the object. Shape descriptors are important when classifying and retrieving objects from databases. Inner and outer complexity, obtained from mutual information calculation with uniformly distributed lines, can be used to classify different families of 2D and 3D objects.
Algorithms of image processing, such as split-and-merge segmentation and image registration, will be presented as paradigmatic examples to understand the basic concepts of entropy, mutual information, data processing inequality, and information bottleneck method.
Mateu Sbert is a full Professor at Department of Information and applied mathematics and the head of the Graphics and Imaging Laboratory, University of Girona, Spain.
His Research Areas include the application of Monte Carlo, Integral Geometry and Information Theory techniques to Video Games, Medical Imaging, Global Illumination, Computational Aesthetics, Visualization, Shape Recognition and Viewpoint Selection. He's currently leading a research team of more than 20 people, Graphics and Imaging laboratory, Gilab(gilab.udg.edu), which holds several quality certifications from Catalan Government.
Mateu Sbert has authored or coauthored more than 150 papers in peer reviewed international journals or conferences, 2 books published by Morgan & Claypool San Francisco, has participated during several years in international program committees of more than 10 different international conferences, including main Eurographics conference and Eurographics Rendering symposium, is currently associated editor of two international journals and has more than 1000 references to his papers. He has led many Spanish research projects, participated in bilateral collaboration projects with several European countries, Canada and China and coordinated VI Framework project Gametools (www.gametools.org). He's a coauthor of 2007 Eurographics tutorial on the subject, and of Courses in Visweek2011 and Siggraph Asia 2011 respectively.