Fast Estimation of Gaussian Mixture Models for Image Segmentation

利用高斯混合模型实现图像分割TheExpectation-Maximizationalgorithmhasbeenclassicallyusedtofindthemaximumlikelihoodestimatesofparametersinprobabilisticmodelswithunobserveddata,forinstance,mixturemodels.Akeyissueinsuchproblemsisthechoiceofthemodelcomplexity.Thehigherthenumberofcomponentsinthemixture,theh