| With the rapid development of computer network, multimedia technology and information hiding technology, the problem of copyright protection of digital works and the integrity certification becomes far more important. Although much has been done on digital watermarking in the past few years, many problems remain to be resolved. In this thesis, we describe two novel digital watermarking algorithms based on the theory of stochastic resonance.First, we present a digital image multiple watermarking algorithm based on stochastic resonance. We map the multi-watermark messages into binary pulse amplitude modulated signals with different code rates, and then add the multi-watermark massages to different DCT coefficients of an image in medium frequency domain, which have been disarranged by Arnold transform. According to the characteristic of multi-watermark, we employ a non-linear bistable detector to detect the multi-watermark massages. This kind of non-linear detector can detect watermarks with different code rates, and is robust for conditions of multi-rate input signals. The experimental results show that this algorithm can resist against several attacks, such as JPEG compression, cut attack, adding Gaussian noise and so on.Furthermore, we research the application of array-noise-enhanced technology in the domain of digital watermarking. A regular network, i.e. an uncoupled parallel array of bistable detectors, can be optimized by non-zero array noise as well as detector parameters. This is the effect of array stochastic resonance, or named by array-noise-enhanced technology. We embed watermark into the original image using improved digital watermarking algorithm. Interestingly, higher similarity between the original character and the extracted one can be further achieved by a parallel array of bistable detectors via the noise-enhanced technology. The present digital watermarking algorithms and related results extend the application of complex random systems to information processing. |
PDF Full Text Request