ReFlowSET: Representation-Aligned Latent Flow Matching for SAR-to-EO Image Translation

Researchers introduced ReFlowSET, a representation-aligned latent flow matching method for SAR-to-EO image translation that improves reconstruction fidelity across codecs and modalities by aligning latent representations. The method addresses limitations of existing latent diffusion approaches that inherit predetermined autoencoders.

SAR-to-EO image translation aims to generate electro-optical EO imagery from synthetic aperture radar SAR observations. Existing latent diffusion approaches typically inherit a predetermined autoencoder, although reconstruction fidelity can vary substantially across codecs and modalities. Becaus