Scholarlink

<p>The growing use of digital imaging in fields like healthcare and cloud storage requires secure and privacy-preserving authentication methods. Traditional approaches often risk data exposure by requiring access to the original image. To address this, we propose a framework that generates hash sequences directly from encrypted images without decryption, ensuring confidentiality. This method applies the Walsh-Hadamard Transform (WHT) and histogram-based feature extraction while leveraging the Paillier cryptosystem’s homomorphic properties for secure computation. The generated hash sequences remain highly robust against compression, noise, and scaling, ensuring reliable authentication even under distortions The proposed method achieves a shorter hash length of 301 bits in 0.12&#xa0;s, thereby reducing the hash generation time when compared to Robust Perceptual Invariant Vector Distance (RPIVD) and Robust Perceptual Nonlinear Matrix Factorization (RPNMF). This reduction makes the proposed approach 20% and 40% faster in hash generation while producing 31.59% and 24.75% shorter hashes compared to RPIVD and RPNMF, respectively. The framework is well-suited for security-critical 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

Efficient image authentication using Walsh-Hadamard Transform in encrypted domain

  • 期刊论文

  • 发表日期：

    2025年7月9日

查看全文
  • S. Apsara,
  • G. Suchitra,
  • A. Meena Kowshalya

摘要

The growing use of digital imaging in fields like healthcare and cloud storage requires secure and privacy-preserving authentication methods. Traditional approaches often risk data exposure by requiring access to the original image. To address this, we propose a framework that generates hash sequences directly from encrypted images without decryption, ensuring confidentiality. This method applies the Walsh-Hadamard Transform (WHT) and histogram-based feature extraction while leveraging the Paillier cryptosystem’s homomorphic properties for secure computation. The generated hash sequences remain highly robust against compression, noise, and scaling, ensuring reliable authentication even under distortions The proposed method achieves a shorter hash length of 301 bits in 0.12 s, thereby reducing the hash generation time when compared to Robust Perceptual Invariant Vector Distance (RPIVD) and Robust Perceptual Nonlinear Matrix Factorization (RPNMF). This reduction makes the proposed approach 20% and 40% faster in hash generation while producing 31.59% and 24.75% shorter hashes compared to RPIVD and RPNMF, respectively. The framework is well-suited for security-critical environments.