Triplet Markov chain in images segmentation

Over the last years, image segmentation has evolved from a sub-discipline of computer science to a technique widely used in medical imaging, automated object recognition, and remote sensing. In this work, we present a recently Markovian model of image segmentation called Triplet Markov Chain with Independent Noise (TMC-IN), in this model, it assumes that its hidden process X is non-stationary. TMC-IN is used in this to segment some textured grey level and color images. To estimate the parameters, we use the iterative algorithm EM (Expectation-Maximization) and we apply MPM (Marginal Posteriori Mode) algorithm to estimate the result segmented image. In addition, we compare the obtained results by this model with those obtained by the stationary Hidden Markov Chain with Independent Noise (HMC-IN) model. Experimental results show that TMC-IN outperforms HMC-IN in all experiments.