Scholarlink

<p>The current research proposes a new zero watermarking approach that aims to protect medical images from tampering by utilizing the powerful learning capabilities of deep learning. In this paper, we propose a robust zero watermarking scheme based on multi-module fusion of VGG16, which enhances the efficiency and security of medical image transmission. The scheme leverages depthwise separable convolution to speed up feature extraction, enabling the focus on global image features. Furthermore, by incorporating the Convolutional Attention Module with channel and spatial attention, the extracted features are enhanced with stronger perceptual, discriminative, and generalization abilities.To preserve multi-scale information and improve feature reusability, jump-joins are employed. The multi-module fusion of VGG16 facilitates the extraction of deep features with superior generalization, multi-scale, and high perceptual capabilities. Binary feature vectors are generated using a mean hash algorithm and encrypted using a novel dual chaos dual logistics encryption method designed in this paper, which significantly enhances both image watermark security and overall network transmission security.The effectiveness of the scheme is verified through extensive experiments. Experimental results demonstrate that the proposed method achieves an NC value of over 0.92 in the presence of noise filtering and rotation attacks, indicating robust security and strong resistance to common image distortions. These results confirm the scheme’s capability to provide high security for medical image transmission, ensuring both robustness against attacks and confidentiality in medical systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Zero Watermarking Algorithms for Medical Images Based on Multi-module Fusion with VGG16 and Dual-Logistic Mapping Encryption

  • 期刊论文

  • 发表日期：

    2025年4月21日

查看全文
  • Jingyou Li,
  • Rongle Wei,
  • Xiaotian Xi,
  • Guangda Zhang,
  • Zixin Yang,
  • Fengshan Zhang

摘要

The current research proposes a new zero watermarking approach that aims to protect medical images from tampering by utilizing the powerful learning capabilities of deep learning. In this paper, we propose a robust zero watermarking scheme based on multi-module fusion of VGG16, which enhances the efficiency and security of medical image transmission. The scheme leverages depthwise separable convolution to speed up feature extraction, enabling the focus on global image features. Furthermore, by incorporating the Convolutional Attention Module with channel and spatial attention, the extracted features are enhanced with stronger perceptual, discriminative, and generalization abilities.To preserve multi-scale information and improve feature reusability, jump-joins are employed. The multi-module fusion of VGG16 facilitates the extraction of deep features with superior generalization, multi-scale, and high perceptual capabilities. Binary feature vectors are generated using a mean hash algorithm and encrypted using a novel dual chaos dual logistics encryption method designed in this paper, which significantly enhances both image watermark security and overall network transmission security.The effectiveness of the scheme is verified through extensive experiments. Experimental results demonstrate that the proposed method achieves an NC value of over 0.92 in the presence of noise filtering and rotation attacks, indicating robust security and strong resistance to common image distortions. These results confirm the scheme’s capability to provide high security for medical image transmission, ensuring both robustness against attacks and confidentiality in medical systems.