A likelihood function for block-based motion analysis.
Pekka SangiJanne HeikkiläOlli SilvénPublished in: ICIP (1) (2005)
Keyphrases
- motion analysis
- likelihood function
- maximum likelihood
- particle filter
- em algorithm
- motion model
- image sequences
- posterior probability
- gaussian process
- expectation maximization
- generative model
- motion compensation
- motion estimation
- parameter space
- point correspondences
- model selection
- motion parameters
- computer vision
- apparent motion
- regression model
- feature vectors
- flow field
- hyperparameters
- machine learning
- maximum a posteriori
- probabilistic model
- viewpoint
- bayesian inference
- computational complexity
- face recognition