Single-Image HDR Reconstruction by Learning to Reverse the Camera Pipeline

被引:174
作者
Liu, Yu-Lun [1 ,2 ]
Lai, Wei-Sheng [3 ]
Chen, Yu-Sheng [1 ]
Kao, Yi-Lung [1 ]
Yang, Ming-Hsuan [3 ,4 ]
Chuang, Yung-Yu [1 ]
Huang, Jia-Bin [5 ]
机构
[1] Natl Taiwan Univ, Taipei, Taiwan
[2] MediaTek Inc, Hsinchu, Taiwan
[3] Google, Mountain View, CA 94043 USA
[4] UC Merced, Merced, CA USA
[5] Virginia Tech, Blacksburg, VA USA
来源
2020 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR) | 2020年
基金
美国国家科学基金会;
关键词
D O I
10.1109/CVPR42600.2020.00172
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Recovering a high dynamic range (HDR) image from a single low dynamic range (LDR) input image is challenging due to missing details in under-/over-exposed regions caused by quantization and saturation of camera sensors. In contrast to existing learning-based methods, our core idea is to incorporate the domain knowledge of the LDR image formation pipeline into our model. We model the HDR-to-LDR image formation pipeline as the (1) dynamic range clipping, (2) non-linear mapping from a camera response function, and (3) quantization. We then propose to learn three specialized CNNs to reverse these steps. By decomposing the problem into specific sub-tasks, we impose effective physical constraints to facilitate the training of individual sub-networks. Finally, we jointly fine-tune the entire model end-to-end to reduce error accumulation. With extensive quantitative and qualitative experiments on diverse image datasets, we demonstrate that the proposed method performs favorably against state-of-the-art single-image HDR reconstruction algorithms.
引用
收藏
页码:1648 / 1657
页数:10
相关论文
共 50 条
[41]   Single-Image Defogging Algorithm Based on Deep Learning [J].
Zhao Jiantang .
LASER & OPTOELECTRONICS PROGRESS, 2019, 56 (11)
[42]   SINGLE-IMAGE STEREOGRAMS [J].
CRONIN, D .
DR DOBBS JOURNAL, 1995, 20 (07) :18-&
[43]   Single image-based HDR image generation with camera response function estimation [J].
Huo, Yongqing ;
Zhang, Xudong .
IET IMAGE PROCESSING, 2017, 11 (12) :1317-1324
[44]   An Attention-Based Network for Single Image HDR Reconstruction [J].
Dafaallah, Mohamed ;
Yuan, Hui ;
Jiang, Shiqi ;
Yang, Ye .
2022 IEEE INTERNATIONAL SYMPOSIUM ON CIRCUITS AND SYSTEMS (ISCAS 22), 2022,
[45]   HDR-DANet: single HDR image reconstruction via dual attentionHDR-DANet: single HDR image reconstruction via dual attentionJ. Ma, H. Zhang [J].
Jindong Ma ;
Haitao Zhang .
Multimedia Systems, 2025, 31 (1)
[46]   Field evaluation of a novel holographic single-image depth reconstruction sensor [J].
Hartlieb, Simon ;
Schober, Christian ;
Haist, Tobias ;
Reichelt, Stephan .
JOURNAL OF THE EUROPEAN OPTICAL SOCIETY-RAPID PUBLICATIONS, 2023, 19 (01)
[47]   Review of Single-Image 3D Face Reconstruction Methods [J].
Wang, Jingting ;
Li, Huibin .
Computer Engineering and Applications, 2023, 59 (17) :1-21
[48]   Single-image human mesh reconstruction by parallel spatial feature aggregation [J].
Li, Shasha ;
Liu, Yuehu ;
Su, Wei .
COMPUTER ANIMATION AND VIRTUAL WORLDS, 2022, 33 (3-4)
[49]   EXTERNAL AND INTERNAL LEARNING FOR SINGLE-IMAGE SUPER-RESOLUTION [J].
Wang, Shuang ;
Lin, Shaopeng ;
Liang, Xuefeng ;
Yue, Bo ;
Jiao, Licheng .
2015 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP), 2015, :128-132
[50]   Single-image super-resolution via local learning [J].
Yi Tang ;
Pingkun Yan ;
Yuan Yuan ;
Xuelong Li .
International Journal of Machine Learning and Cybernetics, 2011, 2 :15-23