MARU-Net: Multiscale Attention Gated Residual U-Net With Contrastive Loss for SAR-Optical Image Matching

Accurate synthetic aperture radar-optical matching is essential for combining the complementary information from two sensors. However, main challenge overcoming different heterogeneous characteristics of imaging In this article, we propose an end-to-end machine learning pipeline inspired by recent advances in image segmentation. We develop a siamese multiscale attention-gated residual U-Net feature extraction satellite images. The architecture shares weights and transforms images into homogeneous space. Fast Fourier transform used to compute cross-correlation between maps produce similarity map. A contrastive loss introduced aid training procedure model maximize discriminability model. experimental results on benchmark dataset show that proposed method has superior accuracy precision compared other state-of-the-art methods.