Bayesian Mixture Models on Connected Components for Newspaper Article Segmentation.
Giorgos SfikasGeorgios LouloudisNikolaos StamatopoulosBasilis GatosPublished in: DocEng (2016)
Keyphrases
- connected components
- mixture model
- level set
- binary images
- maximum likelihood
- gaussian mixture model
- watershed transformation
- em algorithm
- generalized em algorithm
- connected component analysis
- generative model
- dirichlet prior
- probabilistic model
- density estimation
- expectation maximization
- model selection
- image segmentation
- bayesian information criterion
- automatic model selection
- unsupervised learning
- probability density function
- mixture modeling
- segmentation method
- segmentation algorithm
- background subtraction
- connected component labeling
- language model
- bayesian networks
- text lines
- multiscale
- unsupervised image segmentation
- learning algorithm
- raster scan
- image processing