Multi-Domain Rapid Enhancement Networks for Underwater Images
Images captured during marine engineering operations suffer from color distortion and low contrast. Underwater image enhancement helps to alleviate these problems. Many deep learning models can infer multi-source data, where images with different perspectives exist from multiple sources. To this end...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-11-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/23/21/8983 |
| _version_ | 1797631241865920512 |
|---|---|
| author | Longgang Zhao Seok-Won Lee |
| author_facet | Longgang Zhao Seok-Won Lee |
| author_sort | Longgang Zhao |
| collection | DOAJ |
| description | Images captured during marine engineering operations suffer from color distortion and low contrast. Underwater image enhancement helps to alleviate these problems. Many deep learning models can infer multi-source data, where images with different perspectives exist from multiple sources. To this end, we propose a multichannel deep convolutional neural network (MDCNN) linked to a VGG that can target multi-source (multi-domain) underwater image enhancement. The designed MDCNN feeds data from different domains into separate channels and implements parameters by linking VGGs, which improves the domain adaptation of the model. In addition, to optimize performance, multi-domain image perception loss functions, multilabel soft edge loss for specific image enhancement tasks, pixel-level loss, and external monitoring loss for edge sharpness preprocessing are proposed. These loss functions are set to effectively enhance the structural and textural similarity of underwater images. A series of qualitative and quantitative experiments demonstrate that our model is superior to the state-of-the-art Shallow UWnet in terms of UIQM, and the performance evaluation conducted on different datasets increased by 0.11 on average. |
| first_indexed | 2024-03-11T11:20:04Z |
| format | Article |
| id | doaj.art-14b12e0a642b451a8737f7dc65040f33 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-11T11:20:04Z |
| publishDate | 2023-11-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-14b12e0a642b451a8737f7dc65040f332023-11-10T15:13:01ZengMDPI AGSensors1424-82202023-11-012321898310.3390/s23218983Multi-Domain Rapid Enhancement Networks for Underwater ImagesLonggang Zhao0Seok-Won Lee1The Knowledge-Intensive Software Engineering (NiSE) Research Group, Department of Artificial Intelligence, Ajou University, Suwon City 16499, Republic of KoreaThe Knowledge-Intensive Software Engineering (NiSE) Research Group, Department of Artificial Intelligence, Ajou University, Suwon City 16499, Republic of KoreaImages captured during marine engineering operations suffer from color distortion and low contrast. Underwater image enhancement helps to alleviate these problems. Many deep learning models can infer multi-source data, where images with different perspectives exist from multiple sources. To this end, we propose a multichannel deep convolutional neural network (MDCNN) linked to a VGG that can target multi-source (multi-domain) underwater image enhancement. The designed MDCNN feeds data from different domains into separate channels and implements parameters by linking VGGs, which improves the domain adaptation of the model. In addition, to optimize performance, multi-domain image perception loss functions, multilabel soft edge loss for specific image enhancement tasks, pixel-level loss, and external monitoring loss for edge sharpness preprocessing are proposed. These loss functions are set to effectively enhance the structural and textural similarity of underwater images. A series of qualitative and quantitative experiments demonstrate that our model is superior to the state-of-the-art Shallow UWnet in terms of UIQM, and the performance evaluation conducted on different datasets increased by 0.11 on average.https://www.mdpi.com/1424-8220/23/21/8983underwater image enhancementmulti-domain machine learningDCNNdomain adaptabilityperceptual loss |
| spellingShingle | Longgang Zhao Seok-Won Lee Multi-Domain Rapid Enhancement Networks for Underwater Images Sensors underwater image enhancement multi-domain machine learning DCNN domain adaptability perceptual loss |
| title | Multi-Domain Rapid Enhancement Networks for Underwater Images |
| title_full | Multi-Domain Rapid Enhancement Networks for Underwater Images |
| title_fullStr | Multi-Domain Rapid Enhancement Networks for Underwater Images |
| title_full_unstemmed | Multi-Domain Rapid Enhancement Networks for Underwater Images |
| title_short | Multi-Domain Rapid Enhancement Networks for Underwater Images |
| title_sort | multi domain rapid enhancement networks for underwater images |
| topic | underwater image enhancement multi-domain machine learning DCNN domain adaptability perceptual loss |
| url | https://www.mdpi.com/1424-8220/23/21/8983 |
| work_keys_str_mv | AT longgangzhao multidomainrapidenhancementnetworksforunderwaterimages AT seokwonlee multidomainrapidenhancementnetworksforunderwaterimages |