Enhanced Security and Robustness in Color Image Watermarking Using DWT-HD and Quaternion SVD under Hybrid Attacks

Abstract

Securing color images through watermarking presents significant challenges, particularly when subjected to complex hybrid attacks. The objective of this study is to enhance both the security and robustness of Color Image Watermarking (CIW) by developing a novel method that integrates Discrete Wavelet Transform (DWT) and Quaternion Singular Value Decomposition (QSVD) with Heisenberg Decomposition (HD). This combined approach, referred to as DWT-QSVD-HD, has been tested rigorously under various hybrid attack scenarios. The method employs a dual-key strategy, utilizing a logo watermark and the HH component of the R-level DWT, to enhance robustness. A modified watermark insertion rule is proposed, leveraging both keys simultaneously within the HD-QSVD domain, which significantly complicates unauthorized decoding efforts. The LAB color space is utilized during the pre-processing stage to maximize entropy, with entropy analysis providing justification for this adaptation. To further bolster security, the watermarked image is encrypted using a simplified fast AES algorithm, adding an additional layer of protection and improving the watermark's resilience. The performance of the proposed method is evaluated against existing techniques using parametric analysis, optimized via the Fruit Fly Optimization Algorithm (FOA) for key metrics such as Normalized Correlation (NC), Peak Signal-to-Noise Ratio (PSNR), and Structural Similarity Index Measure (SSIM). The method demonstrates exceptional performance across various image sizes, maintaining an NC above 0.995 for 64×64 images. Notably, under motion blur attacks, the proposed method improves the NC from 0.8322 to 0.9003 for 256×256 images. The impact of individual and hybrid watermark attacks is systematically assessed, with results showing superior extraction and recovery of image quality when using the proposed technique. [ABSTRACT FROM AUTHOR]