Fast-MVSNet: Sparse-to-Dense Multi-View Stereo With Learned Propagation and Gauss-Newton Refinement.
Zehao YuShenghua GaoPublished in: CoRR (2020)
Keyphrases
- image features
- multi view stereo
- gauss newton
- scene flow
- camera parameters
- multi view
- depth map
- bundle adjustment
- urban environments
- structure from motion
- d scene
- object recognition
- active appearance models
- feature vectors
- point cloud
- depth estimation
- single view
- three dimensional
- image pairs
- optical flow
- gaussian process
- visual hull
- scene understanding
- stereo correspondence
- high fidelity
- sparse representation
- multiple views
- camera calibration
- stereo matching
- surface reconstruction
- motion field