DRHDR: A Dual Branch Residual Network for Multi-Bracket High Dynamic Range Imaging

Juan Marín-Vega, Michael Sloth, Peter Schneider-Kamp, Richard Röttger; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2022, pp. 844-852

Abstract


We introduce DRHDR, a Dual branch Residual Convolutional Neural Network for Multi-Bracket HDR Imaging. In order to address the challenges of fusing multiple brackets from dynamic scenes, we propose an efficient dual branch network that operates on two different resolutions. The full resolution branch uses a Deformable Convolutional Block to align features and retain high-frequency details. A low resolution branch with a Spatial Attention Block aims to attend wanted areas from the non-reference brackets, and suppress displaced features that could incur on ghosting artifacts. By using a dual branch approach we are able to achieve high quality results while constraining the computational resources required to estimate the HDR results.

Related Material


[pdf]
[bibtex]
@InProceedings{Marin-Vega_2022_CVPR, author = {Mar{\'\i}n-Vega, Juan and Sloth, Michael and Schneider-Kamp, Peter and R\"ottger, Richard}, title = {DRHDR: A Dual Branch Residual Network for Multi-Bracket High Dynamic Range Imaging}, booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops}, month = {June}, year = {2022}, pages = {844-852} }