We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Locally optimum watermark decoder based on fast quaternion generic polar complex exponential transform.
- Authors
Zhang, Si-yu; Wang, Chun-peng; Sun, Yao-ru; Yang, Jun; Gao, Shi-qing
- Abstract
Digital image watermarking, as an important image content security protection technology, has higher research value. It is well-known that imperceptibility, robustness, and watermark payload are three paramount factors to evaluate model performance. Recent methods based on statistical modeling strategy effectively trade-off between imperceptibility and robustness. However, capacity and time complexity are not allowed to ignore. Furthermore, how to select robust modeling objects, appropriate statistical models, and decision rules is one of the major issues in statistical watermark detection. To this end, we propose a color image watermark detector in robust fast quaternion generic polar complex exponential transform (FQGPCET) magnitudes domain, which fully considers the human visual system (HVS) and the statistical properties of image signals. Specifically, we adopt the Cauchy-Rayleigh distribution to model the probability density function of the FQGPCET magnitudes. Then, local high entropy image blocks are used as a new embedding domain for watermark messages due to their strong robustness. Furthermore, we develop a novel optimum multibit watermark decoder based on maximum likelihood theory. Experimental results are evaluated employing a set of standard color images in terms of watermark capacity, imperceptibility, and robustness. The proposed scheme provides better performance against all types of attacks in comparison with other existing methods.
- Subjects
DIGITAL image watermarking; DIGITAL watermarking; WATERMARKS; QUATERNIONS; TIME complexity; DIGITAL images
- Publication
Multimedia Tools & Applications, 2024, Vol 83, Issue 18, p54279
- ISSN
1380-7501
- Publication type
Article
- DOI
10.1007/s11042-023-17682-y