Direct 3-D shape recovery from image sequence based on multi-scale Bayesian network.
Norio TagawaJunya KawaguchiShoichi NaganumaKan OkuboPublished in: ICPR (2008)
Keyphrases
- multiscale
- bayesian networks
- image sequences
- optic flow
- video sequences
- spatio temporal
- diagnostic reasoning
- scale space
- bayesian network inference
- inference in bayesian networks
- conditional independence
- probabilistic inference
- coarse to fine
- structure from motion
- image representation
- image segmentation
- consecutive images
- moving objects
- bayesian network classifiers
- motion estimation
- natural images
- random variables
- graphical models
- background subtraction
- structural learning
- feature points
- parameter learning
- image frames
- image features
- edge detection
- motion analysis
- causal relationships
- learning bayesian networks
- optical flow
- uncertain knowledge
- incomplete data
- apparent motion
- motion detection
- human walking
- motion segmentation