Research On Compressive Sensing Theory And Its Application In Digital Watermarking Technology

Compressive Sensing shows that the sampling rate can be much lower than the limit of the sampling theorem to achieve accurate reconstruction of sparse or compressible signals.Which reduced the information acquisition, storage, processing and transmission cost greatly.The theory is widely used, but it is not mature enough and needs further research.With the rapid development of information technology, the security of reproduction is suffering an increasing number of threats, traditional encryption measures have been unable to solve all sorts of problems that we are facing today. As a branch of information hiding technology,digital watermarking technology has an underestimated role in terms of copyright protection.The existing digital watermarking algorithm research is mostly concentrated in the transform domain, such embedding method has higher security and stronger robustness with respect to airspace algorithm. But its calculation is complex, and the ability against the attack and anti-extraction capacity remains limited. While many features of Compressive Sensing can make up for some shortcomings of digital watermarking technology, mainly as follows:Different sparse domains reflected as different kinds of transform domains in the watermark,which greatly expands the watermark embedding space; Different measurement matrices act as different keys, increasing the security of digital watermark. This paper deeply studies these two kinds of theories, and puts forward a novel digital watermarking algorithm based on the combination of the two. This paper mainly completed the following work:Firstly, this paper introduces three core issues that Compressive Sensing theory mainly related to: the sparse of signal, the construction of the measurement matrix and the reconstruction of the signal. In this paper, we study Compressive Sensing theory from the above three aspects respectively, and then compared with five classical greedy reconstruction algorithms.Secondly, aiming at the limitation of the discrete wavelet transform, which has only horizontal, vertical and diagonal three directions of decomposition, we analyzed Contourlet wave, making full use of the characteristics of multi-directional, multi-resolution to sparse image. And for the limitation of traditional block decision, we study the block theory and adopt interleaving extraction method to reduce the block effect. Then proposed a Contourlet domain Compressive Sensing algorithm based on interleaving extraction block. After simulation, the peak signal to noise ratio of the proposed algorithm has been greatly improved,and it has general applicability.Thirdly, this paper applies Compressive Sensing theory to digital watermarking technology, which mainly manifested in two aspects: the watermark CS preprocess and embedding watermark in CS domain.The simulation results indicate that, it is difficult to extract or destroy the watermark information in both of the two methods when measurement matrix and other information are unknown, verifying the method is feasible, and robust to most attacks.Finally, this paper proposed a Contourlet domain digital watermarking algorithm based on the interleaving extraction block Compressive Sensing. Experimental results demonstrate that the proposed algorithm has better robustness, invisibility and security, and improves the anti-attack ability and anti-extract ability of the image.