Introduction of EM Algorithm into Color Image Segmentation

A color image segmentation method is proposed on the basis of the maximum likelihood (ML) estimation. The observed color image is considered as a mixture of multi-variate normal densities and the number of densities is assumed to be known. The EM algorithm is introduced in order to estimate and improve the parameters of the mixture of densities recursively. The initial parameters for the EM algorithm are estimated by the multidimensional histogram and the minimum distance clustering methods. After the parameter estimation, the segmented image is obtained by the conventional ML method. Consequently, since no random selection is used for initial parameter estimation, the proposed method is stable and useful for unsupervised image segmentation applications. The performance of the algorithm is demonstrated by real color image segmentation experiments.