A Novel Method of Ship Detection in High-Resolution SAR Images Based on GAN and HMRF Models

This research proposes a novel method based on generative adversarial network (GAN) and hidden Markov random field (HMRF) models, for use in large-scale high-resolution synthetic aperture radar (SAR) images. The method consists of three stages. In the first stage, a virtual target and a SAR image are generated by using the GAN model, according to the statistical and gray-level features of the original SAR image used in detection. In the second stage, the virtual target is embedded in the generated image. In the third stage, real targets are detected in the generated image by using the HMRF model. The experiment results show that the proposed algorithm based on GAN and HMRF models can be applied to ship detection in high-resolution SAR images, with high accuracy and processing speed.