Study Of Compressed Sensing Algorithm Of Watermark Image

In order to improve the transmission performance and robustness of the watermarking algorithm, the application of compressed sensing in digital watermarking is studied theoretically and experimentally. Firstly, based on the theory of geometric multiresolution analysis, the contourlet transform is used instead of the traditional multi wavelet transform, and the gray image is obtained by the transform and the low frequency part is obtained.Secondly, according to the sparsity of the high frequency part, using compressed sensing samp algorithm for lossy compression, compression of high frequency coefficient matrix compressed. for gray images, the low frequency part proposed a fast and robust zero watermark algorithm. Zero watermarking algorithm based on bit plane theory to analysis of low frequency image into different bit plane level, bit plane matrix structure-has no value, combined with each plane of the non zero number to generate the image feature matrix. Thirdly, the feature matrix is partitioned into blocks, and the maximum singular value matrix of the block is generated by singular value decomposition, and the zero watermark information is obtained by the chaotic encryption of the generation matrix. Finally, after the zero watermark information extraction, compression after high fre-quency part and the low frequency part of the portfolio by contourlet inverse transform image compression is obtained. Finally, robust detection and transmission performance test. Simulation results show that in watermark robustness, the proposed algorithm has higher robustness, in similar algorithms against salt and pepper noise attack is increased by 6%, against common attacks mixed increased by 12%.in transmission performance, the compressed sensing of gray image transmission efficiency provided 7.5% higher.