BlurDM: A Blur Diffusion Model for Image Deblurring
By: Jin-Ting He , Fu-Jen Tsai , Yan-Tsung Peng and more
Potential Business Impact:
Fixes blurry pictures by reversing how they got blurry.
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.
Similar Papers
Warm Diffusion: Recipe for Blur-Noise Mixture Diffusion Models
CV and Pattern Recognition
Makes AI create better pictures by mixing blur and noise.
Residual-based Efficient Bidirectional Diffusion Model for Image Dehazing and Haze Generation
CV and Pattern Recognition
Turns hazy pictures clear, and clear ones hazy.
MoTDiff: High-resolution Motion Trajectory estimation from a single blurred image using Diffusion models
CV and Pattern Recognition
Makes blurry photos show exact movement paths.