Abstract: This paper proposes an unsupervised change-detection method considering the spatial-contextual information in the SAR log-ratio image.The expectation maximization(EM) algorithm is employed to automatically estimate the statistical parameters associated with changed and unchanged pixels,and the Markov random fields(MRF's) model is applied to characterize the contexture-dependent information.The experiment results show that the proposed method can improve accuracy and reliability of change-detection process.