Simultaneous Bayesian inference of motion velocity fields and probabilistic models in successive video-frames described by spatio-temporal MRFs
Yuya InagakiJun-ichi InouePublished in: CoRR (2010)
Keyphrases
- bayesian inference
- probabilistic model
- video frames
- velocity field
- spatial and temporal
- spatio temporal
- image sequences
- video sequences
- optical flow
- consecutive frames
- vector field
- potential functions
- motion field
- video data
- input video
- graphical models
- conditional random fields
- space time
- fluid motion
- moving objects
- markov random field
- hidden variables
- random fields
- image velocity
- generative model
- flow field
- human motion
- latent variables
- bayesian networks
- motion trajectories
- motion analysis
- motion patterns
- ego motion
- optical flow estimation
- motion segmentation
- motion model
- belief propagation
- dynamic scenes
- motion estimation
- graph cuts
- motion parameters
- higher order
- expectation maximization
- camera motion
- feature points
- three dimensional
- optic flow
- machine learning
- maximum entropy
- prior information
- particle filter
- model selection