Self-supervised Non-uniform Kernel Estimation with Flow-based Motion Prior

for Blind Image Deblurring

Zhenxuan Fang1    Fangfang Wu1*    Weisheng Dong1    Xin Li2    Jinjian Wu1    Guangming Shi1

1Xidian University      2 West Virginia University

 


Figure 1. Overview of the proposed UFPNet for blind image deblurring. The architecture of (a) the flow-based uncertain kernel

estimation network, (b) the encoder-decoder deblurring network with kernel attention module, (c) the kernel attention module (KAM).

 

 

Abstract

Many deep learning-based solutions to blind image deblurring estimate the blur representation and reconstruct the target image from its blurry observation. However, these methods suffer from severe performance degradation in real-world scenarios because they ignore important prior information about motion blur (e.g., real-world motion blur is diverse and spatially varying). Some methods have attempted to explicitly estimate non-uniform blur kernels by CNNs, but accurate estimation is still challenging due to the lack of ground truth about spatially varying blur kernels in real-world images. To address these issues, we propose to represent the field of motion blur kernels in a latent space by normalizing flows, and design CNNs to predict the latent codes instead of motion kernels. To further improve the accuracy and robustness of non-uniform kernel estimation, we introduce uncertainty learning into the process of estimating latent codes and propose a multi-scale kernel attention module to better integrate image features with estimated kernels. Extensive experimental results, especially on real-world blur datasets, demonstrate that our method achieves state-of-the-art results in terms of both subjective and objective quality as well as excellent generalization performance for non-uniform image deblurring.

 

 

 

Paper


                                                                               CVPR 2023                                     Supplementary Material

 

Citation

Zhenxuan Fang, Fangfang Wu, Weisheng Dong, Xin Li, Jinjian Wu, and Guangming Shi. Self-supervised Non-uniform Kernel Estimation with Flow-based Motion Prior

for Blind Image Deblurring. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023.

 

 

Bibtex

@inproceedings{fang2022self,

         title={Self-supervised Non-uniform Kernel Estimation with Flow-based Motion Prior for Blind Image Deblurring },

   author={Fang, Zhenxuan and Wu Fangfang and Dong, Weisheng and Li, Xin and Wu, Jinjian and Shi, Guangming},

   booktitle={ Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},  

   year={2023},

}

 

 

 

 

Download


 

                            

                                                                                                             Github                

 

 

Results

Quantitative Results:

Table 1. The comparison results on the benchmark datasets, the models are trained only on the GoPro dataset.

 


 

 

Visualization Results:

 

Figure 2. Visual comparisons on the GoPro dataset. The estimated kernel at the indicated pixel is illustrated on the left-top.

 


 

Figure 3. Visual comparisons on the RealBlur-J dataset. The estimated kernel at the indicated pixel is illustrated on the left-top.

 

 


 

 

 

 

Contact

Zhenxuan Fang, Email: zxfang@stu.xidian.edu.cn

Fangfang Wu, Email: wufangfang@xidian.edu.cn

Weisheng Dong, Email: wsdong@mail.xidian.edu.cn

Xin Li, Email: xin.li@mail.wvu.edu

Jinjian Wu, Email: jinjian.wu@mail.xidian.edu.cn

Guangming Shi, Email: gmshi@xidian.edu.cn