Online Multibody Factorization Based on Bayesian Principal Component Analysis of Gaussian Mixture Models.
Kentarou HitomiTakashi BandoNaoki FukayaKazushi IkedaTomohiro ShibataPublished in: ICONIP (1) (2008)
Keyphrases
- multibody
- gaussian mixture model
- principal component analysis
- motion segmentation
- maximum likelihood
- mixture model
- motion tracking
- geometric properties
- feature space
- measurement matrix
- structure from motion
- motion model
- speaker recognition
- point correspondences
- background subtraction
- articulated objects
- factorization method
- dynamic scenes
- expectation maximization
- feature vectors
- em algorithm
- rigid objects
- singular value decomposition
- geometric constraints
- speaker identification
- image sequences
- feature tracking
- motion analysis
- model selection
- dimensionality reduction
- face images
- bayesian networks
- face recognition
- mel frequency cepstral coefficients
- posterior distribution
- covariance matrix
- iterative algorithms
- motion estimation
- degrees of freedom
- feature extraction
- three dimensional