Research Of Digital Watermark Algorithm Based On Compressive Sensing

With the popularization of internet and multimedia technology. The storage, reproduction and transmission of digital products are more and more convenient. At the same time, the security of multimedia data suffered more. The development of input and output devices and the emergence of high precision products make the infringement easier. The publishing, transmission and management of confidential information is easy to attract the attention of the attacker because of the emergence of the hacker. This makes the encrypted data as dangerous as the unencrypted data. The traditional encryption measures have been unable to solve all sorts of problems that faces. As a branch of information hiding technology, digital watermarking possess the role to be reckoned with in copyright protection, integrity, authentication and secure communication. Most existing research achievements of digital watermarking algorithms are in transform domain; in comparison with spatial domain, its advantages are larger data volume, higher security and stronger robustness. But its limitations are also obvious:complex computing requirement, weak in resisting attack and anti-extraction.Compressive Sensing(CS) is a new theory proposed by Candes etal. which is superior to the traditional Shannon sampling theory. CS is aimed at sparse or compressible signal, it can finish the signal sampling and data compression at the same time. And the sampling rate is far less than Nyquist sampling theory. CS can reconstruct signal perfectly through a small amount of sampling data by certain optimization algorithm. The sparse of signal, the built of perception matrix and the reconstruction of signal are core processes of CS theory. It has the advantages that DWT and DCT cannot reach of. So in this paper, we apply CS in the research of digital watermarking algorithm to get more robust and secure digital watermark.In this paper, we research the previous digital watermark algorithm and CS theory and the method of reconstruction. Then we propose an improved digital watermarking algorithm which performs digital watermark embedded process in CS domain based on the characteristics of CS and Human eye Visual System(HVS).We select the sub-block with larger capacity by using the non-uniformity of the block to embed the scrambled digital watermark and the algorithm can choose suitable quantization steps adaptively by using quantization method. Experimental results show that the algorithm can ensure the better robustness, concealment and security of digital watermark. At the same time, the ability of defending against attack and extraction of embedded watermark is greatly improved. The watermark is detected without any reference to the original image, which remarkably decreases the storage cost.