Authors: Ruikang Li, Yujin Wang, Shiqi Chen, Fan Zhang, Jinwei Gu, Tianfan Xue

Published on: September 27, 2024

Impact Score: 7.8

Arxiv code: Arxiv:2409.18783

Summary

• What is new: A new dual-domain denoising method called DualDn, which outperforms existing single-domain denoising approaches.
• Why this is important: Existing image denoising methods in the raw or sRGB domains struggle with specific noise distributions and ISP configurations.
• What the research proposes: DualDn employs two denoising networks operating in both the raw and sRGB domains, connected with a differentiable ISP, and adapts to various noise types and ISP settings.
• Results: DualDn achieves state-of-the-art performance, enhances generalizability, and outperforms commercial on-camera denoising solutions.

Technical Details

Technological frameworks used: Learning-based dual-domain denoising with differentiable ISP.

Models used: Two separate denoising networks for raw and sRGB domains.

Data used: Not specified in the abstract.

Potential Impact

Camera manufacturers and companies developing ISP pipelines could benefit from integrating DualDn, potentially disrupting current commercial camera denoising solutions.

Want to implement this idea in a business?

We have generated a startup concept here: DualClear.