The present invention belongs to the image segmentation technical field and relates to a Markov model-based synthetic aperture radar image fast segmentation method. The influence of superpixels and neighborhood systems on segmentation speed is comprehensively researched on the basis of the Markov model, so that the synthetic aperture radar image fast segmentation method is put forward. According to the method of the invention, superpixel preprocessing is performed on a radar image, so that neighborhood systems that can reasonably describe positional relationships can be generated; and on the basis of an MRF (Markov Random Field Model)-based image segmentation and corresponding post-processing, a final segmentation result can be obtained. As proved by experiments, the method can greatly improve image segmentation speed and preserve segmentation quality to the greatest extent.