1. NCCU Academic Hub
  2. Publications
  3. College of Informatics
  4. Proceedings
  5. Item

Publications-Proceedings

Article View/Open

Publication Export

Google ScholarTM

NCCU Library

Citation Infomation

Related Publications in TAIR

題名 BlurDM: A Blur Diffusion Model for Image Deblurring
作者 彭彥璁
He, Jin-Ting;Tsai, Fu-Jen;Peng, Yan-Tsung;Chen, Min-Hung;Lin, Chia-Wen;Lin, Yen-Yu
貢獻者 資訊系
日期 2025-12
上傳時間 20-Mar-2026 10:17:09 (UTC+8)
摘要 Diffusion models show promise for dynamic scene deblurring; however, existing studies often fail to leverage the intrinsic nature of the blurring process within diffusion models, limiting their full potential. To address it, we present a Blur Diffusion Model (BlurDM), which seamlessly integrates the blur formation process into diffusion for image deblurring. Observing that motion blur stems from continuous exposure, BlurDM implicitly models the blur formation process through a dual-diffusion forward scheme, diffusing both noise and blur onto a sharp image. During the reverse generation process, we derive a dual denoising and deblurring formulation, enabling BlurDM to recover the sharp image by simultaneously denoising and deblurring, given pure Gaussian noise conditioned on the blurred image as input. Additionally, to efficiently integrate BlurDM into deblurring networks, we perform BlurDM in the latent space, forming a flexible prior generation network for deblurring. Extensive experiments demonstrate that BlurDM significantly and consistently enhances existing deblurring methods on four benchmark datasets. The source code is available at https://github.com/Jin-Ting-He/BlurDM.
關聯 The Thirty-Ninth Annual Conference on Neural Information Processing Systems (NeurIPS), NeurIPS Foundation
資料類型 conference
dc.contributor 資訊系
dc.creator (作者) 彭彥璁
dc.creator (作者) He, Jin-Ting;Tsai, Fu-Jen;Peng, Yan-Tsung;Chen, Min-Hung;Lin, Chia-Wen;Lin, Yen-Yu
dc.date (日期) 2025-12
dc.date.accessioned 20-Mar-2026 10:17:09 (UTC+8)-
dc.date.available 20-Mar-2026 10:17:09 (UTC+8)-
dc.date.issued (上傳時間) 20-Mar-2026 10:17:09 (UTC+8)-
dc.identifier.uri (URI) https://nccur.lib.nccu.edu.tw/handle/140.119/162134-
dc.description.abstract (摘要) Diffusion models show promise for dynamic scene deblurring; however, existing studies often fail to leverage the intrinsic nature of the blurring process within diffusion models, limiting their full potential. To address it, we present a Blur Diffusion Model (BlurDM), which seamlessly integrates the blur formation process into diffusion for image deblurring. Observing that motion blur stems from continuous exposure, BlurDM implicitly models the blur formation process through a dual-diffusion forward scheme, diffusing both noise and blur onto a sharp image. During the reverse generation process, we derive a dual denoising and deblurring formulation, enabling BlurDM to recover the sharp image by simultaneously denoising and deblurring, given pure Gaussian noise conditioned on the blurred image as input. Additionally, to efficiently integrate BlurDM into deblurring networks, we perform BlurDM in the latent space, forming a flexible prior generation network for deblurring. Extensive experiments demonstrate that BlurDM significantly and consistently enhances existing deblurring methods on four benchmark datasets. The source code is available at https://github.com/Jin-Ting-He/BlurDM.
dc.format.extent 1087516 bytes-
dc.format.mimetype application/pdf-
dc.relation (關聯) The Thirty-Ninth Annual Conference on Neural Information Processing Systems (NeurIPS), NeurIPS Foundation
dc.title (題名) BlurDM: A Blur Diffusion Model for Image Deblurring
dc.type (資料類型) conference