Abstract

In this paper, a new robust digital image watermarking algorithm which was based on singular value decomposition (SVD) and discrete wavelet transform (DWT) was proposed and simulated for protecting real property rights. A gray scale logo image, rather than a randomly generated Gaussian noise type watermark, was used as a watermark. Its embedding algorithm hid a watermark LL sub-band blocks in the low–low (LL) and high-high (HH) sub-bands of a target non-overlapping block of the host image by modifying singular values on SVD version of these blocks. A semi-blind watermark extraction was designed to estimate the original coefficients.
Experimental results showed that the proposed scheme made significant improvements in terms of both transparency and robustness and was superior to the existing methods which were considered in this paper.