English

Learning a Reinforced Agent for Flexible Exposure Bracketing Selection

Computer Vision and Pattern Recognition 2020-05-27 v1

Abstract

Automatically selecting exposure bracketing (images exposed differently) is important to obtain a high dynamic range image by using multi-exposure fusion. Unlike previous methods that have many restrictions such as requiring camera response function, sensor noise model, and a stream of preview images with different exposures (not accessible in some scenarios e.g. some mobile applications), we propose a novel deep neural network to automatically select exposure bracketing, named EBSNet, which is sufficiently flexible without having the above restrictions. EBSNet is formulated as a reinforced agent that is trained by maximizing rewards provided by a multi-exposure fusion network (MEFNet). By utilizing the illumination and semantic information extracted from just a single auto-exposure preview image, EBSNet can select an optimal exposure bracketing for multi-exposure fusion. EBSNet and MEFNet can be jointly trained to produce favorable results against recent state-of-the-art approaches. To facilitate future research, we provide a new benchmark dataset for multi-exposure selection and fusion.

Keywords

Cite

@article{arxiv.2005.12536,
  title  = {Learning a Reinforced Agent for Flexible Exposure Bracketing Selection},
  author = {Zhouxia Wang and Jiawei Zhang and Mude Lin and Jiong Wang and Ping Luo and Jimmy Ren},
  journal= {arXiv preprint arXiv:2005.12536},
  year   = {2020}
}

Comments

to be published in CVPR 2020

R2 v1 2026-06-23T15:48:40.810Z