Unsupervised Domain-Specific Deblurring via Disentangled Representations

By Boyu Lu, Jun-Cheng Chen, Ramalingam Chellappa. 2019

153 citations recorded by the source corpus.

Image deblurring aims to restore the latent sharp images from the corresponding blurred ones. In this paper, we present an unsupervised method for domain-specific, single-image deblurring based on disentangled representations. The disentanglement is achieved by splitting the content and blur features in a blurred image using content encoders and blur encoders. We enforce a KL divergence loss to regularize the distribution range of extracted blur attributes such that little content information is contained. Meanwhile, to handle the unpaired training data, a blurring branch and the cycle-consistency loss are added to guarantee that the content structures of the deblurred results match the original images. We also add an adversarial loss on deblurred results to generate visually realistic images and a perceptual loss to further mitigate the artifacts. We perform extensive experiments on the tasks of face and text deblurring using both synthetic datasets and real images, and achieve improved results compared to recent state-of-the-art deblurring methods.

Subfields: Computer Vision and Pattern Recognition, Generative Models

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Ming-Hsuan Yang
2. Luc Van Gool
3. Yulun Zhang
4. Yujiu Yang
5. Weihao Xia
6. Jing-Hao Xue
7. Radu Timofte
8. Lei Zhang

In Generative Models

1. Yujiu Yang
2. Weihao Xia
3. Jing-Hao Xue
4. Yulun Zhang
5. Ming-Hsuan Yang
6. Wei-Sheng Lai
7. Wenqi Ren
8. B. Stenger

Citations in the corpus

• GAN Inversion: A Survey (2021, background): “[79] disentangle the content and blur features from blurred images for image restoration.”
• Domain Fingerprints for No-Reference Image Quality Assessment (2019, background): “For image restoration, Lu et al. [24] disentangle the content and blur features from blurred images.”
• Deep Image Deblurring: A Survey (2022, methodology): “DR UCSDBN [70] Convolution CycleGAN Unsupervised domain-specific deblurring method by disentangling the content and blur features from input images.”
• Generative Adversarial Networks for Face Generation: A Survey (2022, background): “Many approaches [30, 39, 90, 94, 104, 121] proposed to use GANs to resolve this problem by restoring the latent sharp faces from the corresponding blurred ones.”