Abstract
With the increased use of electronic medical records and computer networks, Medical Image Watermarking (MIW) now plays a very important role to preserve integrity and completeness of medical images. As of now, there are no perfect algorithms or solutions for invisible watermarking as there are trade-offs between visibility and robustness. In this study, we explored multiple implementations of image watermarking techniques using Hybrid Approach and Deep-Learning-Approach. The experiments to measure the limitations and robustness were done on a dataset of breast ultrasound images. 18 attacking methods were performed on the encoded images and performance were evaluated using PSNR and NCC. Encoded images were then being transmitted digitally using multiple transmission method to test its robustness against transmission platform. In conclusion, the Deep-Learning Approach of RivaGAN has shown best robustness despite many extreme attacks while the Hybrid Approach of DWT-DCT-SVD shown the best performance in terms of imperceptibility. We reject RivaGAN as the best solution for Medical Image Watermarking despite its robustness as it was created specifically for video invisible watermarking.
Keywords: Invisible Watermarking, DCT, DWT, SVD, RivaGAN
Authors
Wong Yew Lee [1] ; Loh Jia Cheng [2] ; Li Chen Zhen [3] ; Tan Chi Wee [4]
[1][2][3][4] Faculty of Computing and Information Technology, Tunku Abdul Rahman University of Management and Technology, Kuala Lumpur, Malaysia
[1] wongyewlee-wm19@student.tarc.edu.my
Cite Me
Plain Text:
Y.L.Wong, J.C.Loh, C.Z.Li, C.W.Tan, "A Comparative Study on Medical Image Watermarking using Hybrid Approach and RivaGAN," International Conference on Digital Transformation and Applications (ICDXA) 2021, 2021, pp. 217-225, doi: https://doi.org/10.56453/icdxa.2021.1023.
BibTex:
@INPROCEEDINGS{ICDXA202124,
author={Wong, Yew Lee and Loh, Jia Cheng and Li, Chen Zhen and Tan, Chi Wee},
booktitle={International Conference on Digital Transformation and Applications (ICDXA) 2021},
title={A Comparative Study on Medical Image Watermarking using Hybrid Approach and RivaGAN},
year={2021},
volume={},
number={},
pages={217-225},
doi={https://doi.org/10.56453/icdxa.2021.1024}}
