Multi-view object pose estimation from correspondence distributions and epipolar geometry.
Rasmus Laurvig HaugaardThorbjørn Mosekjær IversenPublished in: CoRR (2022)
Keyphrases
- multi view
- pose estimation
- d objects
- epipolar geometry
- point correspondences
- multiple views
- feature points
- fundamental matrix
- image correspondences
- single view
- visual hull
- epipolar lines
- multiple cameras
- human body
- position and orientation
- viewpoint
- depth map
- rigid objects
- object recognition
- image pairs
- camera motion
- bundle adjustment
- range data
- three dimensional
- dynamic scenes
- camera views
- shape from silhouettes
- camera calibration
- motion parameters
- geometric constraints
- body parts
- pose parameters
- curved surfaces
- stereo pair
- feature matching
- image sequences
- computer vision
- photometric stereo
- camera pose
- range images
- semi supervised
- camera positions
- stereo correspondence
- depth images
- surface reconstruction
- focal length
- closed form
- uncalibrated cameras
- shape from shading
- camera parameters
- machine learning