Deep Learning-Based Joint Enhancement for Compound-Degraded Underwater Imagery

Yue Wu; Tingyu Huang; Yiwei Ma; Shichang Sun

doi:10.54097/zd0s5855

Authors

Yue Wu
Tingyu Huang
Yiwei Ma
Shichang Sun

DOI:

https://doi.org/10.54097/zd0s5855

Keywords:

Deep Learning; Image Enhancement; Data Processing.

Abstract

In complex and dynamic underwater environments, different scenarios are often intertwined with multiple image degradation factors, such as color shift, low illumination, and blur. These composite degradation characteristics make enhancement methods tailored to single scenarios struggle to meet comprehensive requirements. This study aims to design a multi-functional underwater image enhancement model capable of adapting to the challenges of various complex scenarios. The model takes into account diverse degradation features and utilizes image data for validation and optimization. The final enhancement effect is evaluated through both visual display of images and quantitative metrics (including PSNR, UCIQE and UIQM).

References

[1]Pham, Nguyen-Hoang-Hai and Sun, Chi-Chia, ”FPGA-Based Single Image Dehaz-ing System Implementation Using Improved Dark Channel Prior,” 2023 IEEE 12th Global Conference on Consumer Electronics (GCCE), pp. 749-750, 2023.

[2]Chen, Peng, Jamet, C´edric, and Liu, Dong, ”LiDAR Remote Sensing for Vertical Distribution of Seawater Optical Properties and Chlorophyll-a From the East China Sea to the South China Sea,” IEEETransactions on Geoscience andRemote Sensing, vol. 60, pp. 1-21, 2022.

[3]Fan, Junfeng, Wang, Xuan, Zhou, Chao, Ou, Yaming, Jing, Fengshui, and Hou, Zengguang, ”Development, Calibration, and Image Processing ofUnderwater Struc-tured Light Vision System: A Survey,” IEEE Transactions on Instrumentation and Measurement, vol. 72, pp. 1-18, 2023.

[4]Tarekegn, Adane Nega, Cheikh, Faouzi Alaya, Ullah, Mohib, Sollesnes, Erik Tobias, Alexandru, Cornelia, Azar, Saeed Nourizadeh, Erol, Erdeniz, and Suciu, George,” Underwater Object Detection using Image Enhancement and Deep Learning Mod-els,” 2023 11th European Workshop on Visual Information Processing (EUVIP), pp. 1-6, 2023.