Color Appearance Models Instructor: Mark Fairchild (Rochester Institute of Technology) Basic and advanced colorimetry: * Fundamentals of colorimetry * CIE tristimulus values * CIE color spaces * Definition of advanced colorimetry Color appearance phenomena: * When basic colorimetry breaks * Visual phenomena * Chromatic adaptation Fundamentals of color appearance modeling: * Chromatic adaptation models * Construction of color appearance models * CIELAB as a simple example The CIECAM97s color appearance model: * History * Formulation * Use * Future revision Testing, implementation, and application of color appearance models: * Experiments, data for model formulation, and evaluation * Implementation considerations * Applications * Device-independent color imaging --------------------------------------------------------- Computer Vision and the Art of Special Effects Instructors: Steve Sullivan (Industrial Light and Magic) Luc Robert (REALVIZ) Irfan Essa (Georgia Institute of Technology) Steve Seitz (University of Washington) Eugene Vendrovsky (Rhythm & Hues) Intro: Computer vision and film production * How an effects studio works * How computer vision algorithms make their way into artists’ hands Computer vision at ILM * Vision techniques used in "Star Wars : Episode I", "Mighty Joe Young", "The Mummy", "Magnolia", "Pearl Harbor", and "A.I." * The gap between theory and practice Computer vision at Rhythm & Hues * Vision techniques used in "Sum of All Fears", "Harry Porter", and "Dr. Doolittle 2" * Perspective of a matchmove artist Developing commercial products with vision algorithms for production * Experience with REALVIZ * Tools: Camera matchmove, image-based modeling, panoramic stitching, sequence retiming Special effects and computer vision education * "Vision for Graphics" course at Univ. Washington * Special effects course at Georgia Tech. * Examples of completed projects Current and future problems * How current algorithms can fall short * Algorithms we’d like to see in the future --------------------------------------------------------- Developing Computer Vision Applications in Windows Instructors: Ross Cutler (Microsoft Research) DirectShow: * Examples: audio/video capture, processing, display, storage * Visual Studio wizard * DirectX Media Objects Direct3D: * 2D and 3D graphics * Image processing functions GDI+ (Graphics Device Interface): * Drawing text and 2D graphics * Enhancements: anti-aliased lines, alpha blending, gradient and texture fills, ... Win32 API: * System level tasks * Process and thread creation, priorities, events, timers, shared memory, file access, bitmaps, window management Vision SDK: * Basic features: video capture, display, image classes * Advanced topics: sequence classes, property lists, self-describing streams, CLAPACK, extending the SDK Real-time applications Intel Pentium III/4 Architecture --------------------------------------------------------- Face Recognition by Humans and Machines: A Tutorial Survey Instructor: Baback Moghaddam (Mitsubishi Electric Research Laboratory) Role of face recognition in human evolution: social interaction and species survival. How and why the primate brain appears to be specialized for this particular visual task: * Role of cortical areas IT and STS * Results from single-cell recordings * Study of performance deficits in humans Survey of computer vision methods for automatic face recognition from the 1970s to the present: * Feature-based, appearance-based, view-based, 2D shape-texture models, 3D shape-texture models * Subspace representations * Matching algorithms Overview of DARPA’s FERET program: * Primary goals * Evaluation methodology * Key findings The current state-of-the-art and future challenges: * Pose, illumination and non-rigid deformation * Pros and cons of 3D vs. 2D view-based modeling and recognition strategies Face recognition as a biometric and its applications in a society with ubiquitous cameras and increasingly automated surveillance --------------------------------------------------------- Image-based Lighting Instructor: Paul Debevec (USC Institute for Creative Technologies) * Introduction * An overview of how light works * Global illumination overview * How camera measure light * Recovering high dynamic images from photos * Recording real-world illumination * Illuminating synthetic objects with real light * Making "Rendering with Natural Light" (SIGGRAPH’98 Electronic Theater) * Rendering synthetic objects into real scenes * Making "Fiat Lux" (SIGGRAPH’99 Electronic Theater) * Image-based lighting with commercial renderers and in commercial production * Overview of real-time image-based lighting techniques * Image-based lighting real objects and actors --------------------------------------------------------- Image Search Engines: Techniques and Applications Instructors: Theo Gevers (University of Amsterdam) Arnold Smeulders (University of Amsterdam) Color, texture, and shape features: * Color survey: standard color models, color invariance * Appropriate use of color models * Overview of shape features for image retrieval Searching: * Improving classification through experience * Survey of classification techniques Indexing: * Making image retrieval more efficient * Survey of indexing methods Visualization: * Visualizing feature matching results * Object localization Relevance feedback: * Learning from feedback * Methods and systems using feedback * Survey of image retrieval system --------------------------------------------------------- Level-set Methods and Partial Differential Equations in Computer Vision and Image Processing Instructors: Stanley Osher (UCLA) Guillermo Sapiro (University of Minnesota) Ron Fedkiw (Stanford University) Fundamentals: * Derivation of level-set equations * Basic numerical implementation * Variational level sets * Fast numerical implementations * Constructing the embedding function * PDE’s on surfaces Applications: * Image segmentation * Image interpolation * Image enhancement * Object tracking * Pattern generation * Vector field denoising and visualization * Image inpainting * Natural phenomena simulation Discussion and the future --------------------------------------------------------- Medical Image Analysis: Deformable Models and Registration Instructor: Herve Delingette (INRIA, Sophia Antipolis, France) Introduction: * The different image modalities * Volumetric images vs. video Image segmentation: * Iconic-based image segmentation: - Thresholding and morphology - The EM algorithm: Brain segmentation from MR images * Model-based image segmentation: - Deformable model segmentation - Simplex meshes - Registration to local deformation - Boundary and region-based external forces - Examples: Liver, heart, ... Image registration: * Geometric (feature-based) registration: - Extraction of geometric features in volumetric images - Algorithms: Tree search, geometric hashing, iterative closest points * Iconic (intensity-based) registration: - Similarity measures - Measure of deformation smoothness - Minimization of dual energy Applications: * Multiple sclerosis evolution * Augmented reality in the operating room * Multimodal image fusion * Atlas superposition --------------------------------------------------------- Multiple View Geometry Instructors: Anders Heyden (Lund University, Sweden) Marc Pollefeys (K.U. Leuven, Belgium) Introduction: * Motivation * Structure and motion problems * Applications Tensor calculus and projective geometry: * Tensor calculus and algebra * Projective geometry * The plane at infinity and the absolute conic Modeling cameras: * The pinhole model * Calibrated and uncalibrated cameras * Calibration Multiple view geometry: * The fundamental matrix * The trifocal tensor * The quadrifocal tensor Structure and motion I: * Using multi-view tensors * Factorization * Auto-calibration Structure and motion II: * Feature extraction * Feature similarity and matching * Structure and motion initialization * Structure and motion updating * Bundle adjustment Flexible calibration: * Self-calibration * Critical motion and the absolute conic * Practical methods Dense depth estimation: * Image pairs rectification * Dense stereo matching * Multi-view depth estimation Visual modeling: * 3D-modelling * Unstructured lightfield rendering Examples and applications Conclusions --------------------------------------------------------- Open Source Computer Vision Library: Overview, Optimization, Algorithms, and Hands-on Use Instructor: Gary R Bradski (Intel Corporation) Victor Eruhimov (Intel Corporation) Vadim Pisarevsky (Intel Corporation) OpenCV contents overview: * Demos Programming and algorithm tutorial: * Calibration Tutorial and Use * USB Stereo Depth Tutorial and Use * Kalman filter, HMM, optical flow * Matlab interface * Interpretive C environment * Run time profiling tool. Distribution of CD set containing: * All Intel performance libraries * OpenCV * Documentation and tutorials on programming with Intel chips Limited version of Vtune profiling code Optional session for users who bring their own laptops: * Choose a programming assignment * Experts will be on hand to: - Install and get OpenCV working on WinOS or Linux - Give one-on-one help for any problems or questions arising while completing assignment