Decoupling Cross-Modality Manifold Discrepancy: Leveraging Visible Diffusion Priors for Infrared Super-Resolution

Researchers propose Shift-IISR, a dual-path diffusion framework for infrared image super-resolution.

It combines global and local guidance to better align distributions and preserve fine details.

The model aims to reduce cross-modality mismatch, producing clearer and more faithful upscaled infrared images.

This could improve downstream sensing, surveillance, and other low-light vision applications.