| Title |
Solving Blind Inverse Problems with Unpaired Data via Optimal Transport and Diffusion Priors |
| Authors |
김민우(Minwoo Kim) ; 임홍기(Hongki Lim) |
| DOI |
https://doi.org/10.5573/ieie.2026.63.1.49 |
| Keywords |
Blind inverse problems; Optimal transport; Diffusion models |
| Abstract |
This paper proposes a novel framework for addressing blind inverse problems solely using unpaired datasets. Conventional approaches often assume prior knowledge of the operator or require paired degraded and clean images. In contrast, our proposed method eliminates such assumptions entirely, making it particularly suitable for real-world scenarios where paired data is unavailable and the operator is unknown. We first estimate the operator by learning an optimal transport map from the clean data distribution to the degraded data distribution. This estimated operator is then integrated into a diffusion-based posterior sampling scheme, which is utilized to solve the inverse problem in a non-blind manner. Experimental results show that the proposed method achieves superior performance compared to existing diffusion-based approaches on blind inverse problems. |