Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces ProSR, a novel approach for high-resolution Synthetic Aperture Radar (SAR) imagery that reformulates image super-resolution as a semantically-guided discrete token prediction task. By leveraging a Self-Supervised Learning backbone and Semantic-Aligned Detail Encoding, ProSR effectively preserves the impulsive scattering characteristics of SAR while mitigating distortions commonly seen in existing diffusion models. Experimental validation on a large-scale benchmark demonstrates that ProSR not only enhances visual quality but also maintains the critical physical consistency required for SAR applications.
ProSR achieves high-resolution SAR imagery that retains essential scattering characteristics, outperforming traditional generative models plagued by structural distortions.
High-resolution Synthetic Aperture Radar (SAR) imagery is critical for precision analysis such as automatic target recognition, yet its acquisition is costly. Although generative image super-resolution (ISR) models offer a promising alternative, current smooth-approximation based diffusion frameworks often struggle to preserve the coherent scattering statistics, causing stochastic structural distortions that are less consistent with real SAR physics. To address this, we propose Semantic Prototype-Guided Super-Resolution (ProSR), reformulating SAR ISR as a semantically-guided discrete token prediction task within a quantized latent space. By mapping signal features to discrete scattering primitives, ProSR preserves the impulsive nature of SAR without over-smoothing. Furthermore, we integrate a Self-Supervised Learning backbone into SAR ISR to extract label-free semantic priors, overcoming label scarcity. Guided by these priors, we introduce Semantic-Aligned Detail Encoding to decouple high-frequency signals into discrete scattering primitives. In parallel, the Semantic Prototype Map Generator explicitly constructs semantic prototype maps, allowing Prototype-Map-Guided Attention to route the information flows within identical categories and mitigate inter-class interference. To validate our approach, we present a large-scale 0.25m resolution benchmark from the Umbra Open Dataset. Experimental results show ProSR achieves superior visual quality while preserving essential scattering characteristics required for practical SAR applications.