DOGAN: DINO-Based Optical-Prior-Driven GAN for SAR-to-Optical Image Translation

To leverage the complementary advantages of synthetic aperture radar’s (SAR) all-weather and all-day imaging capability and optical imagery’s intuitive visualization, SAR-to-optical image translation (S2OIT) has emerged as a promising solution to mitigate the interpretability challenges posed by SAR’s speckle noise and geometric distortions. However, the scale of high-quality registered SAR-optical data is limited, where incorporating priors is a viable solution. What’s more, the digging out of optical prior is insufficient among the existing methods, leading to inadequate synthesis of optical-like texture in translated optical images. To address these challenges, we propose DOGAN, a distillation with no labels (DINO)-based optical-prior-driven generative adversarial network (GAN) framework that integrates ample optical priors extracted from a pretrained DINO model into the S2OIT process. Specifically, to fully exploit the tremendous optical prior preserved in pretrained DINO and extract the multiscale optical prior, a DINO-based optical-prior extraction (DOE) module is proposed. Furthermore, to elevate the domain adaptability of the optical prior, a lightweight stacked optical-aware (SOA) adapter is proposed to finetune DINO for remote sensing data with minimal trainable parameters. To instill the extracted affluent optical prior into the S2OIT pipeline stably, the SAR-optical multiscale domain alignment (SO-MDA) module is proposed, which employs L1 and multikernel maximum mean discrepancy (MK-MMD) losses to align intermediate optical and S2O features. Extensive experiments on SAR2Opt and SEN1-2 datasets demonstrate that DOGAN achieves state-of-the-art performance in both translation fidelity and structural realism. To the best of our knowledge, this is the first work to leverage DINO-based optical priors for the S2OIT task.