Deep Hybrid Camera Deblurring

Under Review

Deep Hybrid Camera Deblurring

Jaesung Rim,     Junyong Lee,     Heemin Yang,     Sunghyun Cho



Mobile cameras, despite their significant advancements, still face low-light challenges due to compact sensors and lenses, leading to longer exposures and motion blur. Traditional solutions like blind deconvolution and learning-based methods often fall short in handling ill-posedness of the deblurring problem. To address this, we propose a novel deblurring framework for multi-camera smartphones, utilizing a hybrid imaging technique. We simultaneously capture a long exposure wide-angle image and ultra-wide burst images from a smartphone, and use the sharp burst to estimate blur kernels for deblurring the wide-angle image. For learning and evaluation of our network, we introduce the HCBlur dataset, which includes pairs of blurry wide-angle and sharp ultra-wide burst images, and their sharp wide-angle counterparts. We extensively evaluate our method, and the result shows the state-of-the-art quality.

The HCBlur Dataset

Hybrid Camera Blur dataset

Google Drive Link

Google drive link (Soon)

Postech Link

Postech link (Soon)


For the hybrid camera deblurring, we introduce a deep network with two sub-networks: the Hybrid Camera Deblurring Network (HC-DNet) and the Fusion Network (HC-FNet). These networks deblur blurry wide-angle images by leveraging the strengths of our dual-camera hybrid imaging system. HC-DNet uses a burst of sharp ultra-wide images to compute optical flows and transform them to per-pixel blur kernels for deblurring the wide-angle image. HC-FNet refines the output of HC-DNet using the entire sequence of ultra-wide burst images to compensate for any residual blurs or kernel estimation errors from HC-DNet.

Results on HCBlur

Deblurring results on the HCBlur dataset.

Methods PSNR / SSIM
MIMO-UNet+ 22.42 / 0.5966
Uformer-B 23.31 / 0.6332
HINet 23.60 / 0.6360
NAFNet-32 23.60 / 0.6363
NAFNet-64 24.18 / 0.6607
RAFT + NAFNet-32 24.39 / 0.6597
HC-DNet 26.20 / 0.7274
HC-DNet + HC-FNet 26.83 / 0.7413

Results on real images

Deblurring results on real-world images.


 title={Deep Hybrid Camera Deblurring},
 author={Rim, Jaesung and Lee, Junyong and Yang, Heemin and Cho, Sunghyun},
 journal={arXiv preprint},

Let's Get In Touch!

Please feel free to contact us with any feedback, questions, or comments