Progressive Learning of a Diffusion-based Inpainting Model for Separating Overlapped Fingerprints

This paper treats overlapped fingerprint separation as an image-inpainting problem. Starting from a pretrained Stable Diffusion model, the authors progressively add a fingerprint prior, partial-print completion, and overlap-aware inpainting with multi-channel conditioning. Experiments on two public datasets reportedly show that reconstructed component fingerprints can match their mated counterparts with very high probability. The abstract does not provide dataset names, quantitative scores, baselines, or details about the matching protocol, so the strength of the claim cannot be independently assessed from the supplied metadata alone.