BlurDM: A Blur Diffusion Model for Image Deblurring

Min-Hung Chen (NVIDIA) · Jin-Ting He (National Yang Ming Chiao Tung University) · Fu-Jen Tsai (National Tsinghua University) · Yan-Tsung Peng (National Chengchi University) · Chia-Wen Lin (National Tsing Hua University) · Yen-Yu Lin (National Yang Ming Chiao Tung University)

benchmark datasetsblur formation processdeep learning architectures diffusion modelsdual denoising formulationdual-diffusion forward schemedynamic scene deblurringflexible prior generation networkgaussian noise conditioninggenerative modeling image restorationlatent space integrationmotion blur modelingnoise diffusionsharp image recovery

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.