Using structure from motion for monument 3D reconstruction from images with heterogeneous background.
Anna D. SergeevaVictoria A. SablinaPublished in: MECO (2018)
Keyphrases
- structure from motion
- reconstruction from multiple images
- multi view stereo
- feature correspondences
- bundle adjustment
- camera positions
- image based localization
- camera pose estimation
- urban scenes
- three dimensional
- image sequences
- pinhole camera
- missing data
- image correspondences
- camera parameters
- computer vision
- point cloud
- calibrated images
- motion stereo
- focal length
- camera pose
- feature tracking
- line correspondences
- geometric information
- camera motion
- feature points
- motion estimates
- input image
- scene depth
- reprojection error
- multibody
- scene structure
- multiple images
- depth map
- geometric features
- motion estimation
- architectural scenes
- machine learning
- projective reconstruction
- geometric constraints
- vanishing points
- position and orientation
- feature matching