Next:
Acknowledgments
Contents
Acknowledgments
Notations
Contents
Introduction
3D from images
Overview
Projective geometry
Introduction
Projective geometry
The projective plane
Projective 3-space
Transformations
Conics and quadrics
The stratification of 3D geometry
Projective stratum
Affine stratum
Metric stratum
Euclidean stratum
Overview of the different strata
Conclusion
Camera model and multiple view geometry
Introduction
The camera model
A simple model
Intrinsic calibration
Camera motion
The projection matrix
Relation between projection matrices and image homographies
Deviations from the camera model
Multi view geometry
Two view geometry
Relation between the fundamental matrix and image homographies
Three view geometry
Multi view geometry
Conclusion
Relating images
Feature point extraction
similarity measures
Intensity cross-correlation
Matching using affinely invariant regions
Two view geometry computation
Eight-point algorithm
Seven-point algorithm
More points...
Robust algorithm
Degenerate case
Three and four view geometry computation
Structure and motion
Initial frame
Initial projection matrices
Initial structure
Adding a view
projective pose estimation
Refining structure
Initialize new structure
Relating to other views
Example
Refining structure and motion
Levenberg-Marquardt minimization
Bundle adjustment
Conclusion
Self-calibration
Calibration
Scene knowledge
Camera knowledge
Self-calibration
A counting argument
Geometric interpretation constraints
The image of the absolute conic
Self-calibration methods
Critical motion sequences
Flexible self-calibration
Linear approach
Non-linear approach
Maximum Likelihood approach
Conclusion
Dense depth estimation
Stereo matching
Exploiting scene constraints
Constrained matching
Image pair rectification
Planar rectification
Polar rectification
Epipolar geometry
Rectification method
Examples
Multi-view stereo
Correspondence Linking Algorithm
Occlusions and visibility
Depth estimation and outlier detection
Some results
Conclusion
Modeling
Surface model
Texture enhancement
Volumetric model
Plenoptic model
structure and motion
Plenoptic modeling and rendering
Regular grid representation
Representation with recorded images
Experiments
conclusion
Augmented reality
Previous Work
Overview
Structure and motion
Augmented Video
Virtual Object Embedding
Virtual Object Merging
Examples
Discussion
Conclusion
Some results
Introduction
Acquisition of 3D models from photographs
A Jain Temple in Ranakpur
The Béguinages of Leuven
Acquisition of 3D models from pre-existing image sequences
Virtualizing archaeological sites
Virtualizing scenes
Reconstructing an overview model
Reconstructions at different scales
Combination with other models
More applications in archaeology
3D stratigraphy
Generating and testing building hypotheses
Applications in other areas
Architecture and heritage conservation
Mars rover control
Other applications
Conclusion
Bibliography
Curriculum Vitae
About this document ...
Marc Pollefeys 2000-07-12
