Deep Gradual Multi-Exposure Fusion Via Recurrent Convolutional Network
- Authors
- Ryu, Je-Ho; Kim, Jong-Han; Kim, Jong-Ok
- Issue Date
- 2021
- Publisher
- IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
- Keywords
- Feature extraction; Image fusion; Fuses; Image restoration; Image reconstruction; Brightness; Deep learning; Multi-exposure image fusion; recurrent convolutional network; dilated convolution filter; gradual fusion
- Citation
- IEEE ACCESS, v.9, pp.144756 - 144767
- Indexed
- SCIE
SCOPUS
- Journal Title
- IEEE ACCESS
- Volume
- 9
- Start Page
- 144756
- End Page
- 144767
- URI
- https://scholar.korea.ac.kr/handle/2021.sw.korea/138698
- DOI
- 10.1109/ACCESS.2021.3122540
- ISSN
- 2169-3536
- Abstract
- The performance of multi-exposure image fusion (MEF) has been recently improved with deep learning techniques but there are still a couple of problems to be overcome. In this paper, we propose a novel MEF network based on recurrent neural network (RNN). Multi-exposure images have different useful information depending on their exposure levels, and in order to fuse them complementarily, we first extract the local detail and global context features of input source images, and both features are separately combined. A weight map is learned from the local features for effectively fusing according to the importance of each source image. Adopting RNN as a backbone network enables gradual fusion, where more inputs result in further improvement of the fusion gradually. Also, information can be transferred to the deeper level of the network. Experimental results show that the proposed method achieves the reduction of fusion artifacts and improves detail restoration performance, compared to conventional methods.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - College of Engineering > School of Electrical Engineering > 1. Journal Articles
![qrcode](https://api.qrserver.com/v1/create-qr-code/?size=55x55&data=https://scholar.korea.ac.kr/handle/2021.sw.korea/138698)
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.