Rubost Digital Audio Watermarking Algorithms

The dissertation focuses mainly on digital audio watermarking basing on the frequency domain. Digital watermarking is to determine the ownership or examine the primitiveness of the digital product by inserting information in it. The interrelated investigation focuses mainly on digital image watermarking and a cross investigation together with other science domain, but less on digital audio watermarking.From the point of view of the digital watermarking system, the dissertation make an improving on pretreatment and blind detecting based on halftone technology and independent component analysis.The main content of the dissertation are summed up as follows.In part 1, the objective of the dissertation is introduced, and so as to the significance of research on Digital Watermarking, and the actual investigational status.In part 2, the basic concept of digital watermarking and digital audio watermarking is summarized, and the relation of the both. The dissertation also summarized the classic algorithm, attacking means, and the standard of evaluation. The classic algorithm of digital audio watermarking is introduced in two parts: time domain algorithm and the frequency domain of digital audio watermarking algorithm, the analysis of both are provided.In part 3, the basic concept of halftone image binary theory, and the dissertation bring forward a digital audio watermarking in wavelet domain based on halftone pretreatment. Compared with the traditional digital watermarking, the watermarking is adopted with pseudo noise (PN) sequences or simple binary image (such as the brand). The applicable range of the watermarking technology is confined in a parochial area because of little information for better hiding. It is a first step to a wide area that the digital watermarking technology is applied in secrecy communication, but the hidden information is more than the traditional one. The algorithm in this dissertation inserts a pretreated high altitude picture into audio carrier. The objective of the algorithm is not the robust improved but the split the difference between the hiding effect and the quantity of information. The quantity of pretreated information is 1/8 of the original one, but the maininformation is saved. The result of the experiment indicates that the secrecy communication can be carried out.In part 4, the basic concept of Independent Component Analysis is summarized, and the dissertation put forward a blind detecting watermarking algorithm based on ICA. The traditional one based on quantizing coefficients is less robust to vicious attack. Independent Component Analysis (ICA) is to seek a linear transform for a random variable in order to make each component as possible as independent reciprocally. The watermarking information is blindly detected from mixed signal source by received carrier. The experiment is divided into two parts, (1) an image watermarking scheme in Contour-let domain based on ICA; (2) an audio watermarking scheme in Wavelet domain based on ICA. The result of the experiment indicates that the digital watermarking algorithm based ICA is robust to compression and noise attack.In part 5, a summary to the full text and a plan for next research are given.