Research On Digital Watermarking Algorithm Based On The Compressed Sensing

Since the rapid development of Internet technology, a large amount of digital multimedia content is distributing on the web. Due to the digital product is easy to copy, modify, storage and disseminate through the Internet have made copyright protection of digital property a challenging problem. As the primary technology for the protection of copyright, digital watermarking obtained the widespread attention in recent years. Based on Compressed sensing theory, the digital watermarking with grey images is studied in this thesis which the main content are as follows:(1) We propose a novel compressed sensing based digital watermarking algorithm. Firstly, the Wavelet transform had been made to the image watermark, and got the sparse signal of the watermark. Then a scrambled block Hadamard ensemble matrix structured as a sensing matrix to sense the sparse signal of the watermark to obtain the measurements, and the sensing matrix was saved as key. The Discrete cosine transform had been made to the original image, and the Discrete coefficients obtained was divided into blocks where the measurements of the watermark should be put into. Experiments show that this method for watermark has well invisibility as well as strong robustness.(2) A novel zero-watermarking algorithm using the compressed sensing feature proposed. The zero-watermarking basic theory is introduced at first. Then, we begin to research the algorithm, original image is divided into blocks, and structure a scrambled block Hadamard ensemble matrix as a sensing matrix to sensing each block to obtain the compressed sensing significant feature. Secend, each pixel value of the zero-watermark is generated by modify the value of the compressed sensing significant feature. At last zero-watermarking is registered, and the sensing matrix is stored as the key. The experimental results indicated that this algorithm has a great strong robustness and the original image without any distortion, so proposed algorithm has a wide practical prospect.