Maximum likelihood identification of multiscale stochastic models using the wavelet transform and the EM algorithm.
Vassilios V. DigalakisKenneth C. ChouPublished in: ICASSP (4) (1993)
Keyphrases
- em algorithm
- maximum likelihood
- wavelet transform
- stochastic models
- multiscale
- expectation maximization
- mixture model
- stochastic processes
- stochastic model
- wavelet coefficients
- likelihood function
- parameter estimation
- gaussian mixture model
- maximum likelihood estimation
- subband
- image compression
- wavelet domain
- generative model
- multiresolution
- marginal distributions
- maximum a posteriori
- hyperparameters
- expectation maximisation
- log likelihood function
- image processing
- hidden variables
- image segmentation
- incomplete data
- gaussian distribution
- feature extraction
- penalized likelihood