A multiresolution EM algorithm for unsupervised image classification.
Jean-Marc LafertéFabrice HeitzPatrick PérezPublished in: ICPR (1996)
Keyphrases
- em algorithm
- image classification
- multiresolution
- expectation maximization
- finite mixture model
- maximum likelihood
- unsupervised learning
- mixture model
- maximum likelihood estimation
- parameter estimation
- likelihood function
- image representation
- image features
- gaussian mixture model
- generative model
- gaussian mixture
- expectation maximisation
- minimum message length
- incomplete data
- feature extraction
- wavelet transform
- likelihood maximization
- log likelihood
- semi supervised
- subband
- visual words
- density estimation
- probabilistic principal component analysis
- finite mixtures
- probabilistic model
- supervised learning
- hidden variables
- probability density function
- maximum a posteriori
- trend analysis
- mixture distribution
- model selection
- machine learning
- penalized likelihood
- wavelet coefficients