Morphological wavelet domain image watermarking

Current digital watermarking methods based on the discrete wavelet transform use orthogonal wavelet kernels. In this paper, we discuss the use of morphological wavelets, which are a class of non-linear wavelets, in digital watermarking for scalable coded images. Three different scenarios for embedding the watermark, namely, 1) embedding only in the low pass subband (low-low), 2) embedding only in the high pass subbands (low-high, high-low and high-high) and 3) embedding in all subbands are considered to model popular wavelet domain watermarking methods. The performance of morphological Haar and higher length median wavelets in terms of embedding and detection under content adaptation attacks, such as resolution and quality scalable decoding are shown. Morphological wavelets based watermarking shows a high robustness against the content adaptation attacks, especially in resolution scalability, compared to the conventional orthogonal wavelets based watermarking.