Compression Awareness And Its Application In Digital Watermarking

Compressive Sensing is an emerging field of signal processing theory. The article applied the theory to digital watermark, analyzed color image watermark algorithm which based on compressive sensing to reach the possibility of improving digital watermark algorithms’ robustness and security on compressive sensing.The first part of the article introduced the compressive sensing. Analyze the compressive sensing for its research background and research status at the beginning. Then, mako a brief exposition of the basic principles and theoretical framework of compressive sensing, and analyzed and summarized three key processes of compressive sensing like sparse representation, random measurement and reconstruction algorithms. At last, verify the theory can be applied to the reconstructed image by experiments, studied the effect of image reconstruction based on orthogonal wavelet transform compressive sensing and get the final conclusions through comparing the finally experiment results. Experiments show that compressive sensing can be finely used for image reconstruction; in the specific sparse representation the differen sample rates can lead to different effects of the image reconstruction.The second part of paper conducted a brief overview of the digital watermark, first introduce the steganography technology, which leads the digital watermark technology, then summarize the following aspects like the basic model and the concept of digital watermark system, the digital watermark feature, digital watermark classification, digital watermark have several important technical, such as generating the watermark signal, typical digital watermark embedding method, the digital watermark attacks, digital watermark detection and extraction. Through the introduction we have a certain understanding for some related concepts of digital watermark technology.Finally, combine the compressive sensing theory with digital watermark technology to analyze and compare the experiment of RGB color image watermark algorithm based on compressive sensing theory. In order to have a comprehensive and stereoscopic image of this theory, we first introduce the relevant theoretical knowledge and experiments related basis. In this part, we introduce the theories used in experiments (such as the Haar wavelet basis functions, the traditional Arnold scrambling, traditional evaluation methods of watermark system) and the basic experiment’s method (such as the watermark embedding process, the watermark embedding position selected and watermark pretreatment process).After having a certain understanding of the theoretical basis for the experiments, specifically describe the experiments’method steps and analysis of final results and conclusions. We selected a color images as vector images in the experiments, the watermark image is a homemade binary image, according to the characteristics that the human visual system HVS model of the human eye is not very sensitive to changes in the blue layer of the image, the watermark will be embedded into the blue layer of the image to ensure that vector image has a larger embedding capacity to improve the robustness of the watermark. Selecting discrete wavelet transform method as watermark embedding method, wavelet function select more commonly used Haar wavelets. Before the watermark embedded in the carrier image, usually have a preprocessing to the watermark image, we use compressive sensing method in preprocessing the watermark, the step of compressive sensing methods are consistent with earlier experiments, select the appropriate sampling rate based on the results of the foregoing. Finally, the experimental result that the algorithm takes advantage of the sparsity and adjustable compression ratio of compressive sensing, controls the embedding capacity of the watermark information while improving the security of the watermark. And compare analysis with traditional Arnold scrambling preprocessing, the experiment data obtained shows the algorithm has good robustness under common attacks.