Unsupervised SAR Image Segmentation Based on Triplet Markov Fields With Graph Cuts

Abstract

The triplet Markov fields (TMF) model is suitable for dealing with nonstationary synthetic aperture radar (SAR) images. Existing optimization approaches for the TMF model cannot balance segmentation accuracy and computational efficiency. Focusing on efficient optimization of the TMF model, we propose an unsupervised SAR image segmentation algorithm based on TMF with graph cuts (GCs) in this letter. Considering the existence of two label fields in the TMF model, an iterative optimization strategy under the criterion of maximum a posteriori is proposed, which iteratively estimates one label field with the other fixed. GCs are is used to find the optimal estimation of each label field. GCs optimization and parameter estimation using iterative conditional estimation perform iteratively, leading to an unsupervised segmentation algorithm. Experiments on simulated and real SAR images demonstrate that the proposed algorithm can obtain accurate segmentation results with reasonable computational cost.