Study On Technology Of Digital Audio Watermarking

In recent years, the rapid development of computer and Internet technology has changed our life a lot. It is beneficial for us to access or spread the digital multimedia works via the Internet and rework them easily with all kinds of software, which brings the problem of security and ownership protection.As a new and effective solution, digital watermarking is widely attention-getting. Digital watermarking is a branch of data hiding, which protects digital works by embedding unperceivable and secret information into the works and then extracting the information according to given algorithms. Generally, imperceptibility, robustness and security are the three basic characters of watermarking. Many researches are carried on to optimize these characters now. Image and video watermarking technologies have been carried on widely in recent years. As a new technology, audio watermarking is becoming a hot topic gradually because of its application perspective.The development, status, basic principle and framework of digital watermarking are summarized in this thesis. Then the basic system, applications, classifications and metewand of audio watermarking are introduced. The thesis also summarizes the existing audio watermarking and attacking algorithms. Then based the previous works, research work on audio watermarking based on DWT are carried on, aiming at the improvement of the imperceptibility and robustness of the blind watermarking algorithms:①Considering the character of human audio system, this thesis proposes a blind audio watermarking algorithm in wavelet domain based on Toral scramble and masking effect. Pretreatment to the watermarking image with Toral transform is given first. Initial audio information should be transformed by DWT, then the watermarking information are embedded according to the rule based on non-simultaneous masking effect. Experimental results show that the proposed algorithm has a good imperceptibility and it is a blind watermarking. The algorithm is robust to most common signal processing attacks, but weak at resisting de-synchronization attacks.②Because most audio watermarking algorithms presented in DWT domain are not obvious to resist de-synchronization attacks, according to the improvement upon the first algorithm, another blind watermarking algorithm which is effective to resist de-synchronization attacks is proposed. The error-correcting is applied to the pretreatment of the watermarking information in order to lower the bit error rate and improve the robustness. The proposed algorithm embeds the watermarking information in the absolute mean of low frequency sub-band coefficients in order to get a strong robustness because the mean of signal is not sensitive to de-synchronization attacks. The extraction of the watermarking is blind. Experimental results show that the scheme not only can resist the common signal processing attacks better but also has a strong robustness to common de-synchronization attacks.