Abstract: A new segmentation algorithm is proposed for polarimetric SAR images based on Markov random field combined with Freeman decomposition and scattering entropy.Firstly,Freeman decomposition theory is adopted to obtain four kinds of scattering characteristics,which are surface scattering,double scattering,volume scattering and mixed scattering,and scattering entropy is exacted by H/Alpha decomposition.Then,the terrain can be divided into 12 initial clusters.Secondly,an agglomerative hierarchical clustering algorithm is introduced to reduce the number of clusters.Finally,by adopting Wishart distribution to model the character field,simulated annealing model method(SA) based on maximum a posteriori criterion is used to acquire the segmentation results.Using fully polarimetric SAR images acquired by the NASA/ JPL and E-SAR,the experimental results show the effectiveness and superiority of this algorithm.