Provably-Convergent Iterative Methods for Projective Structure from Motion.
Shyjan MahamudMartial HebertYasuhiro OmoriJean PoncePublished in: CVPR (1) (2001)
Keyphrases
- structure from motion
- iterative methods
- provably convergent
- reconstruction from multiple views
- shape from shading
- bundle adjustment
- epipolar constraint
- image sequences
- projective reconstruction
- missing data
- camera parameters
- computationally expensive
- image reconstruction from projections
- camera motion
- point cloud
- computer vision
- feature tracking
- scene structure
- reconstruction from multiple images
- constrained optimization
- reprojection error
- multiview geometry
- focal length
- iterative algorithms
- optical flow
- three dimensional
- shape recovery
- neural model
- perspective projection
- neural network
- autocalibration
- line segments
- computationally efficient
- motion estimation
- machine learning