Deep Learning for HDR Imaging: State-of-the-Art and Future Trends

  title={Deep Learning for HDR Imaging: State-of-the-Art and Future Trends},
  author={Lin Wang and Kuk-Jin Yoon},
  journal={IEEE transactions on pattern analysis and machine intelligence},
  • Lin Wang, Kuk-Jin Yoon
  • Published 20 October 2021
  • Computer Science
  • IEEE transactions on pattern analysis and machine intelligence
High dynamic range (HDR) imaging is a technique to allow a greater dynamic range of exposures, which is a very important field in image processing, computer graphics, and vision. Recent years have witnessed a striking advancement of HDR imaging using deep learning. This paper aims to provide a systematic review and analysis of the recent development of deep HDR imaging methodologies. Overall, we hierarchically and structurally group existing deep HDR imaging methods into five categories based… 
HDR Reconstruction from Bracketed Exposures and Events
This paper presents a multi-modal end-to-end learning-based HDR imaging system that fuses bracketed images and event modalities in the feature domain using attention and multi-scale spatial alignment modules and proposes a novel event- to-image feature distillation module that learns to translate event features into the image-feature space with self-supervision.
Multi-Bracket High Dynamic Range Imaging with Event Cameras
This paper proposes the first multi-bracket HDR pipeline combining a standard camera with an event camera, and shows better overall robustness when using events, with improvements in PSNR by up to 5dB on synthetic data and up to 0.7dB on real-world data.
Self-supervised HDR Imaging from Motion and Exposure Cues
Experimental results show that the HDR models trained using the proposed self-supervision approach achieve performance competitive with those trained under full supervision, and are to a large extent superior to previous methods that equally do not require any supervision.


Multi-Scale Dense Networks for Deep High Dynamic Range Imaging
A novel deep convolutional neural network is proposed to generate HDR, which attempts to produce more vivid images by using the coarse-to-fine scheme to gradually reconstruct the HDR image with the multi-scale architecture and residual network.
Transfer Deep Learning for Reconfigurable Snapshot HDR Imaging Using Coded Masks
This paper proposes a joint design for snapshot HDR imaging by devising a spatially varying modulation mask in the hardware and building a deep learning algorithm to reconstruct the HDR image, and demonstrates that the proposed hardware–software so lution offers a flexible yet robust way to modulate per‐pixel exposures.
Deep high dynamic range imaging of dynamic scenes
A convolutional neural network is used as the learning model and three different system architectures are compared to model the HDR merge process to demonstrate the performance of the system by producing high-quality HDR images from a set of three LDR images.
Deep Optics for Single-Shot High-Dynamic-Range Imaging
This work fabricates an optimized optical element and attaches it as a hardware add-on to a conventional camera during inference and demonstrates that this end-to-end deep optical imaging approach to single-shot HDR imaging outperforms both purely CNN-based approaches and other PSF engineering approaches.
Deep Multi-Stage Learning for HDR With Large Object Motions
This paper proposes to split the HDR problem into multiple stages and tackle them using separate Convolutional Neural Networks (CNNs) rather than attempting an end-to-end learning.
NTIRE 2021 Challenge on High Dynamic Range Imaging: Dataset, Methods and Results
This paper reviews the first challenge on high-dynamic range (HDR) imaging that was part of the New Trends in Image Restoration and Enhancement (NTIRE) workshop, held in conjunction with CVPR 2021.
Deep HDR Reconstruction of Dynamic Scenes
This paper puts forward an approach combining both traditional pipeline and deep learning algorithm, in which the input LDR images are aligned to the reference one using an optical flow method based on deep convolution network and then merged into HDR images using a powerful network which is called MergeNet.
HDR image reconstruction from a single exposure using deep CNNs
This paper addresses the problem of predicting information that have been lost in saturated image areas, in order to enable HDR reconstruction from a single exposure, and proposes a deep convolutional neural network (CNN) that is specifically designed taking into account the challenges in predicting HDR values.