SVD-based watermarking algorithm is one of the most preferable algorithms for copyright protection due to its singular values (SVs) that have outstanding stability and represents intrinsic algebraic image properties. Hence, there is a good trade-off between robustness and imperceptibility. However, most SVD-based algorithms have been tested against conventional attacks, such as image manipulations, that do not fully exploit adversary’s knowledge. These algorithms are vulnerable to false-positive problem, where an adversary’s watermark can be detected in the watermarked image although it was never inserted. The underlying problem is due to the strong influence of

$U$

and

$V$

orthogonal vectors of SVD on an image. In order to solve false-positive problem,

$UV$

can be used to embed the watermark together with SVs. However, this solution is not ideal as

$U$

and

$V$

hold important structural information of an image and is hypersensitivity to even a little change in

$UV$

vectors. Therefore, this research has the objectives to analyse the robustness of existing SVD-based watermarking algorithms that are using orthogonal vectors and then propose a new robust algorithm that is able to solve false-positive problem and sensitivity issued caused by scaling factor. Hence, a new transform domain image watermarking scheme that utilized both

$U$

and

$V$

orthogonal vectors is proposed with the usage of Human Visual System (HVS), Discrete Wavelet Transform (DWT) and SVD. Experimental results showed that the proposed scheme is more robust against majority types of image processing and geometrical attacks compared to existing schemes while achieving good quality watermarked image level. The significance of new algorithm comes at the right time during the Covid-19 epidemic as organizations involving in business and financial services can be assured of the integrity of its downloadable/streamable/shareable digital files, which are copyrighted through the unique SVD and robustness features of the algorithm ensuring piracy prevention of their content.

Yuan Ju Teoh

Publications