TY - JOUR
T1 - Two-Stage Separable Adversarial Distortion-Based Robust Watermarking Framework for Diffusion Tensor Imaging
AU - Zheng, Long
AU - Li, Zhi
AU - Liu, Zhangyu
AU - Li, Dandan
AU - Zhang, Li
AU - Yue, Hong
AU - Cheng, Fei
AU - Mao, Qin
AU - Wei, Xuekai
AU - Zhou, Mingliang
N1 - Publisher Copyright:
© 2024 World Scientific Publishing Company.
PY - 2024
Y1 - 2024
N2 - Recent deep learning-based watermarking methods have achieved impressive results. However, they struggle with unknown distortions and often suffer from poor generalization, slow convergence, unstable training, and degraded visual quality in watermarked images. To address the above problems, this paper proposes a two-stage separable adversarial distortion (TSAD)-based robust watermarking algorithm for diffusion tensor imaging (DTI). The algorithm uses a noise-free end-to-end network in the first stage for learning and training DTI images. In the second stage, it fixes the watermark embedding network trained in the first stage, interacts the noise distortion network with the watermark extraction network to perform adversarial training for improving robustness. Experimental results show that our method achieves comparable or better robustness to seen distortions and better robustness to unseen distortions, along with enhanced stability, faster convergence, and improved visual quality in watermarked DTI images.
AB - Recent deep learning-based watermarking methods have achieved impressive results. However, they struggle with unknown distortions and often suffer from poor generalization, slow convergence, unstable training, and degraded visual quality in watermarked images. To address the above problems, this paper proposes a two-stage separable adversarial distortion (TSAD)-based robust watermarking algorithm for diffusion tensor imaging (DTI). The algorithm uses a noise-free end-to-end network in the first stage for learning and training DTI images. In the second stage, it fixes the watermark embedding network trained in the first stage, interacts the noise distortion network with the watermark extraction network to perform adversarial training for improving robustness. Experimental results show that our method achieves comparable or better robustness to seen distortions and better robustness to unseen distortions, along with enhanced stability, faster convergence, and improved visual quality in watermarked DTI images.
KW - adversarial training
KW - Deep learning
KW - DTI
KW - robust blind watermarking
UR - http://www.scopus.com/inward/record.url?scp=85196379638&partnerID=8YFLogxK
U2 - 10.1142/S0218001424540119
DO - 10.1142/S0218001424540119
M3 - Article
AN - SCOPUS:85196379638
SN - 0218-0014
VL - 38
JO - International Journal of Pattern Recognition and Artificial Intelligence
JF - International Journal of Pattern Recognition and Artificial Intelligence
IS - 7
M1 - 2454011
ER -