Multiwavelet-based image watermarking using SVD and BP neural network

LI Guo-liang,TAN Yue-hui,XUAN Ke-cheng,LI Zhi-yong,PENG Tao
2010-01-01
Abstract:A blind watermarking algorithm based on CARDBAL multiwavelet transformation,SVD and BP neural network is proposed.In this scheme,the original image is firstly decomposed by multiwavelet transformation,then the relationship among singular values of subblocks in high frequency areas is learned by BP neural network.Finally a watermark signal is embedded into the multiwavelet domain partly by adjusting the relationship among singular values of these subblocks.The experimental results prove the robustness and imperceptibility of this method.
What problem does this paper attempt to address?