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

Jia-Hao Wu,Fu-Jen Tsai,Yan-Tsung Peng, Charles Tsai,Chia-Wen Lin,Yen-Yu Lin

Computing Research Repository (CoRR)（2024）

National Yang Ming Chiao Tung University | National Tsing Hua University | National Chengchi University | Qualcomm Inc

Cited 0|Views21

Abstract

Image deblurring aims to remove undesired blurs from an image captured in adynamic scene. Much research has been dedicated to improving deblurringperformance through model architectural designs. However, there is little workon data augmentation for image deblurring. Since continuous motion causesblurred artifacts during image exposure, we aspire to develop a groundbreakingblur augmentation method to generate diverse blurred images by simulatingmotion trajectories in a continuous space. This paper proposes ImplicitDiffusion-based reBLurring AUgmentation (ID-Blau), utilizing a sharp imagepaired with a controllable blur condition map to produce a correspondingblurred image. We parameterize the blur patterns of a blurred image with theirorientations and magnitudes as a pixel-wise blur condition map to simulatemotion trajectories and implicitly represent them in a continuous space. Bysampling diverse blur conditions, ID-Blau can generate various blurred imagesunseen in the training set. Experimental results demonstrate that ID-Blau canproduce realistic blurred images for training and thus significantly improveperformance for state-of-the-art deblurring models. The source code isavailable at https://github.com/plusgood-steven/ID-Blau.

Translated text

Key words

Image Deblurring,Training Set,Data Augmentation,Architectural Design,Realistic Images,Diverse Images,Motion Trajectory,Continuous Motion,Image Sharpness,Blurred Images,Training Data,Convolutional Neural Network,Diffusion Process,Attention Mechanism,Generative Adversarial Networks,Diffusion Model,Continuous State,Performance Gain,Optical Flow,Stable Strategy,Continuous Field,Noisy Images,Training Pairs,Vision Transformer,Image Synthesis,Complex Training,Real-world Images,Additional Training Data

Bibtex

AI Read Science

AI Summary

AI Summary is the key point extracted automatically understanding the full text of the paper, including the background, methods, results, conclusions, icons and other key content, so that you can get the outline of the paper at a glance.

Example

Background

Key content

Introduction

Methods

Results

Related work

Fund

Key content

Pretraining has recently greatly promoted the development of natural language processing (NLP)
We show that M6 outperforms the baselines in multimodal downstream tasks, and the large M6 with 10 parameters can reach a better performance
We propose a method called M6 that is able to process information of multiple modalities and perform both single-modal and cross-modal understanding and generation
The model is scaled to large model with 10 billion parameters with sophisticated deployment, and the 10 -parameter M6-large is the largest pretrained model in Chinese
Experimental results show that our proposed M6 outperforms the baseline in a number of downstream tasks concerning both single modality and multiple modalities We will continue the pretraining of extremely large models by increasing data to explore the limit of its performance

Try using models to generate summary,it takes about 60s

Must-Reading Tree

Example

Generate MRT to find the research sequence of this paper