Research On Robust Digital Watermarking Based On BSS

With the development of the communication and network technology, more and more digitized information is spreading over the world. Therefore, how to implement effective intellectual copyright protection for these digital products is an emergent problem. Many researchers hence have proposed a new technique, also referred to as digital watermarking, to address this issue.Robustness is not only a core issue of the watermarking technique, but also the research hot topics of the current digital watermarking methods. However, it remains unresolved that how to design such a watermarking algorithm which can resist all kinds of attacks and make the embedded watermark difficult to be removed. Currently, most of robust watermarking algorithms can resist some simple attacks, but can not resist the combined attacks.Based on Bind Source Separation, this paper presents a robust watermarking algorithm, which can resist some simple attacks and some assembled attacks. This method consists of three major steps: 1) when embedding the watermark, it first decomposes the host image using the Tree-structure wavelet, followed by embedding multiple watermark copies into the subdiagram; 2) In the extraction process, the watermarked image is decomposed based on the secret key to obtain the sub-images containing the watermark signal, and then, the Blind Source Separation technique is applied to these sub-images to extract multiple watermark copies; 3) Finally, both the image clustering analysis and image fusion are carried out on these watermark copies so as to obtain the optimal watermark.Since the singled Bind Source Separation method may be ineffective under some circumstances, this paper has improved the performance by incorporating multiple Blind Source Separation schemes. During the period of extraction, multiple Blind Source Separation methods are utilized to extract the watermarks so that the failing possibility of the single Blind Source Separation method can be avoided, thus improving the watermark robustness.The experimental results show that the proposed algorhtms are effective and practical. The shemes can be widely used in e-commerce, e-government, national defense, national security, digital libraries and the field of copyright protection.