Scholarlink

<p>Digital image watermarking is vital for authentication, copyright protection, tamper detection and ownership verification, particularly in IoT environments where the integrity and authenticity of images are crucial. Existing techniques, however, face challenges such as high computational complexity and energy consumption, making them unsuitable for resource-constrained IoT devices. The traditional Singular Value Decomposition (SVD) method, despite its robustness that outperforms competitor techniques in the field, fails to address these challenges effectively. In this paper, we propose a novel image watermarking technique that leverages the SVD technique with a block-based and randomized approach in order to enhance performance and security. By dividing the host image into smaller blocks and using a Pseudo-Random Number Generator (PRNG) with a secret key to select random blocks for embedding, the technique reduces computational complexity and improves robustness against attacks. The watermark is embedded into the singular values of selected blocks in a random way, ensuring a balance between robustness and imperceptibility. This approach effectively addresses the computational and bandwidth constraints of IoT devices while maintaining robustness and high image quality. Evaluation results demonstrate significant performance improvements over the traditional SVD method, underscoring the technique suitability for resource-constrained IoT applications. At 4096 <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\times \)</EquationSource> </InlineEquation> 4096 image resolution, it attains up to 53.5 dB PSNR (Peak-Signal-to-Noise-Ratio) compared to 49.7 dB for classical SVD, while reducing runtime from 10.85 minutes to 4.68 minutes. Furthermore, the method maintains high robustness, with Normalized Correlation (NC) values above 0.98 under common image processing attacks such as Gaussian noise, JPEG compression, motion blur, cropping, and median filtering. These results highlight our proposal suitability for secure, efficient, and imperceptible watermarking in IoT and other resource-limited environments.</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

A randomized block-based SVD image watermarking technique for IoT applications

  • 期刊论文

  • 发表日期：

    2026年1月27日

查看全文
  • Benaidja Amira,
  • Farek Lazhar

摘要

Digital image watermarking is vital for authentication, copyright protection, tamper detection and ownership verification, particularly in IoT environments where the integrity and authenticity of images are crucial. Existing techniques, however, face challenges such as high computational complexity and energy consumption, making them unsuitable for resource-constrained IoT devices. The traditional Singular Value Decomposition (SVD) method, despite its robustness that outperforms competitor techniques in the field, fails to address these challenges effectively. In this paper, we propose a novel image watermarking technique that leverages the SVD technique with a block-based and randomized approach in order to enhance performance and security. By dividing the host image into smaller blocks and using a Pseudo-Random Number Generator (PRNG) with a secret key to select random blocks for embedding, the technique reduces computational complexity and improves robustness against attacks. The watermark is embedded into the singular values of selected blocks in a random way, ensuring a balance between robustness and imperceptibility. This approach effectively addresses the computational and bandwidth constraints of IoT devices while maintaining robustness and high image quality. Evaluation results demonstrate significant performance improvements over the traditional SVD method, underscoring the technique suitability for resource-constrained IoT applications. At 4096 \(\times \) 4096 image resolution, it attains up to 53.5 dB PSNR (Peak-Signal-to-Noise-Ratio) compared to 49.7 dB for classical SVD, while reducing runtime from 10.85 minutes to 4.68 minutes. Furthermore, the method maintains high robustness, with Normalized Correlation (NC) values above 0.98 under common image processing attacks such as Gaussian noise, JPEG compression, motion blur, cropping, and median filtering. These results highlight our proposal suitability for secure, efficient, and imperceptible watermarking in IoT and other resource-limited environments.