A collection of methods for high dynamic range imaging High dynamic range imaging is an important topic on image processing, especially in mobile photograpyh. There are mainly two methods for HDR imaging. One method reconstructs HDR content from a single LDR image. The other is reconstructed from multi-exposure LDR images fusion, which focus on dealing with the problem of ghosting. The recent important studies are summarized here.
HDR image reconstruction from a single exposure using deep CNN(2017 ACM TOG)
Single-image hdr reconstruction by learning to reverse the camera pipeline(2020 CVPR)
Deep optics for single-shot high-dynamic-range imaging(2020 CVPR)
Learning rank-1 diffractive optics for single-shot high dynamic range imaging(2020 CVPR)
Single Exposure High Dynamic Range Image Reconstruction Based on Deep Dual-Branch Network(2021)
Deep high dynamic range imaging of dynamic scenes(2017 SIGGRAPH Aisa)
Deep HDR Reconstruction of Dynamic Scenes(2018 Conference on Image, Vision and Computing)
Deep High Dynamic Range Imaging with Large Foreground Motions(2018 ECCV)
Attention-guided Network for Ghost-free High Dynamic Range Imaging(2019 CVPR)
Deep HDR Video from Sequences with Alternating Exposures(2019)
Multi-scale Dense Networks for Deep High Dynamic Range Imaging(2019 WACV)
Deep HDR imaging via a non-local network(2020)
HDR-GAN HDR Image Reconstruction from Multi-Exposed LDR Images with Large Motions(2020)
Towards Practical and Efficient High-Resolution HDR Deghosting with CNN(2020)
Superpixel-Based Quality Assessment of Multi-Exposure Image Fusion for Both Static and Dynamic Scenes(2021)
UPHDR-GAN: Generative Adversarial Network for High Dynamic Range Imaging with Unpaired Data(2021)
Low-light Image Restoration with Short- and Long-exposure Raw Pairs(2021)