ID-Blau: Image Deblurring by Implicit Diffusion-based reBLurring AUgmentation

TL;DR

ID-Blau addresses the data augmentation gap in image deblurring by introducing an implicit diffusion-based reblurring method. It represents blur as a pixel-wise, continuous blur condition field $C=[x;y;z]$ and conditions a diffusion model on sharp images to synthesize realistic blur $B$, enabling sampling of diverse motion trajectories. The approach yields substantial, consistent improvements when used to augment training for state-of-the-art deblurring models across GoPro, HIDE, and RealBlur datasets, and ablations confirm the benefits of diffusion conditioning, continuous blur fields, and realistic blur modeling. This diffusion-based augmentation offers a practical, controllable, and scalable way to enrich deblurring data, potentially serving as a standard augmentation tool for dynamic-scene restoration tasks. $

Abstract

Image deblurring aims to remove undesired blurs from an image captured in a dynamic scene. Much research has been dedicated to improving deblurring performance through model architectural designs. However, there is little work on data augmentation for image deblurring. Since continuous motion causes blurred artifacts during image exposure, we aspire to develop a groundbreaking blur augmentation method to generate diverse blurred images by simulating motion trajectories in a continuous space. This paper proposes Implicit Diffusion-based reBLurring AUgmentation (ID-Blau), utilizing a sharp image paired with a controllable blur condition map to produce a corresponding blurred image. We parameterize the blur patterns of a blurred image with their orientations and magnitudes as a pixel-wise blur condition map to simulate motion trajectories and implicitly represent them in a continuous space. By sampling diverse blur conditions, ID-Blau can generate various blurred images unseen in the training set. Experimental results demonstrate that ID-Blau can produce realistic blurred images for training and thus significantly improve performance for state-of-the-art deblurring models. The source code is available at https://github.com/plusgood-steven/ID-Blau.

Figures (9)

• Figure 1: Examples of continuous reblurring by ID-Blau, where blur condition maps represent pixel-wise blur information, consisting of blur orientations and magnitudes, in a continuous space. ID-Blau can take a sharp image and a blur condition map to synthesize a blurred image, even unseen in the training set. Condition A is a blur condition map computed from the GoPro training set, which can be used to reblur a sharp image to generate a blurred image as provided in the training set. We can create Condition B and C based on A to synthesize new reblurred images, where Condition B is Condition A with rotated orientations, and Condition C is Condition A added with a camera motion blur.
• Figure 2: Illustration of distributions of blur magnitudes (left) and orientations (right) of the GoPro training set.
• Figure 3: Reblurring process with ID-Blau. As the pie chart shows the blur condition field with orientations and magnitudes delineated in different colors, we visualize the process of generating blurred images with a set of blur condition maps. A sharp image paired with a blur condition map and a noise map is concatenated and fed into ID-Blau to produce a blurred image, where an MLP is used to encode the iteration index $t$ as Time Embedding. Using ID-Blau can augment an image deblurring training set offline for optimizing a deblurring model and improving its performance.
• Figure 4: Illustration of reblurred images with ID-Blau. It takes a sharp image $S$ and a blur condition map $C=[x; y; z]$ as inputs to generate the corresponding blurred image. We show several generated blurred images by altering $C$, such as unit horizontal or vertical blur orientations, $C^1=[1; 0; z]$ and $C^2=[0; 1; z]$, horizontally inverse $C$ as $C^3=[-x; y; z]$, and $C^4=[-x; y; 2z]$ with twice the magnitude for $C^3$.
• Figure 5: Qualitative results on the GoPro testing set (left) and the HIDE dataset (right).