Spatially constrained non-Gaussian mixture model for image segmentation

A new color image segmentation algorithm based on spatial information with the Dirichlet mixture model is presented. This method uses Markov Random Field to incorporate spatial information between neighboring pixels into a Dirichlet mixture model. The segmentation model is learned using Expectation Maximization (EM) algorithm based on Newton Raphson step. The obtained results using real image data set are more encouraging than those obtained using similar approaches.