A. 8:30 - 8:40, 10 min�
� Introduction
��� 1. Overview of area and the course
��� 2. Speaker introductions
B. 8:40 - 9:30, 50 min
� Sensing for vision and graphics (Nayar)
���
����1. The dimensions of visual sensing
��� 2. Catadioptric vision
��� 3. Panoramic and omnidirectional cameras
��� 4. Spherical mosiacs
��� 5. Single camera stereo
��� 6. Radiometric self calibration
��� 7. High dynamic range imaging
��� 8. Vision and the atmosphere
��� 9. Structure from bad weather
C. 9:30 - 10:00, 30 min�
� Overview of passive vision techniques (Seitz)
��� 1. Cues for 3D inference (parallax, shading, focus, texture)
��� 2. Single view techniques
��� 3. Multiple view techniques
������ - Stereo
<> 10:00 - 10:15 Break
D. 10:15 � 10:30, 15 min�
� Overview of passive vision techniques (Seitz)
��� 3. Multiple view techniques
������ - Structure from motion
��� 4. Other approaches
E. 10:30 - 11:15, 45 min�
� Voxels from images (Seitz)
��� 1. Voxel-based scene representation
��� 2. Volume intersection
������ - Shape from silhouettes
��� 3. Voxel coloring
��� 4. Space carving
��� 5. Related techniques
F. 11:15 - 12:00, 45 min�
� Fa�ade: modeling architectural scenes (Debevec)
��� 1. Constrained structure recovery
������ - Architectural primitives
��� 2. Photogrammetry
������ - Recovering camera parameters
������ - Importance of user-interaction
����3. Model-based stereo
��� 4. Connections to image-based rendering
������ - Impact of geometric accuracy on rendering quality
��� ���- Local vs. global 3D models
<> 12:00 - 1:30 Lunch
G. 1:30 - 2:15, 45 min�
� Overview of active vision techniques (Curless)
��� 1. Imaging radar
������ - Time of flight
������ - Amplititude modulation
��� 2. Optical triangulation
������ - Scanning with points and stripes
������ - Spacetime analysis
��� 3. Interferometry
������ - Moire
��� 4. Structured light applied to passive vision
������ - Stereo
������ - Depth from defocus
��� 5. Reflectance capture
������ - From shape-directed lighting
������ - Using additional lighting
H. 2:15 - 3:00, 45 min�
� Desktop 3D photography (Bouguet)
��� 1. Traditional scanning is expensive, but...
�������� desklamp + pencil = structured light
��� 2. The shadow scanning technique
������ - Indoor: on the desktop
������ - Outdoor: the sun as structured light
��� 3. Calibration issues
��� 4. Temporal analysis for improved accuracy
��� 5. Error Analysis
<> 3:00 - 3:15 Break
I. 3:15 - 3:55, 40 min�
� Shape and appearance from images and range data (Curless)
��� 1. Registration
��� 2. Reconstruction from point clouds
��� 3. Reconstruction from range images
������ - Zippering
������ - Volumetric merging
��� 4. Modeling appearance
J. 3:55 - 4:55, 60 min�
� Application: The Digital Michelangelo Project (Levoy)
��� 1. Goals
������ - Capturing the shape and apperance of:
��������� - Michelangelo's sculptures
��������� - Renaissance architecture
��� 2. Motivation
������ - Scholarly inquiry
������ - Preservation through digital archiving
������ - Virtual museums
������ - High fidelity reproductions
��� 3. Design requirements
������ - Geometry: from chisel marks to building facades
�������- Appearance: reflectance of wood, stone, marble
��� 4. Custom scanning hardware
����5. Capturing appearance with high resolution photographs
G. 4:55 - 5:00, 5 min�
� Closing
<> Adjourn