Robust Digital Audio Watermarking Algorithms

With the development of the internet and digital media, people can download audio, video and image works from the internet freely, which is beneficial to our daily life but brings the problem of the ownership protection. Consequently, it becomes more and more important to do research on copyright protection and authentication of the digital media. As an effective solution for protecting audio works, digital audio watermarking technique develops fast in the recent years, however, some issues are still open. The paper focuses on robust audio watermarking algorithms and proposes two robust audio watermarking schemes. Simulation results prove the feasibility of the proposed algorithms. The main contributions are as follows:(1) The meaning of the research, the research background, and the related theoretical concepts are introduced, including audio masking, discrete wavelet transform (DWT), discrete cosine transform (DCT) and blind source separation (BSS).(2) As an effective BSS method, fast independent component analysis (Fast ICA) is used widely in digital audio watermarking, however, it restricts its application due to its time-consuming iterations. Taking the problem aforementioned into consideration, a robust digital audio watermarking algorithm based on independency metric (IM) is proposed. The ICA approach based on IM can be formulated as a generalized eigenvalue (GE) problem, which has no time-consuming iterative learning, compared with traditional BSS methods, such as Fast ICA. The analysis and simulation results prove the robustness, imperceptibility and security of the scheme.(3) It is difficult to control the embedding strength to balance robustness and imperceptibility perfectly in the traditional schemes. Besides, the traditional schemes may distort the original audio to some extent and can not achieve the balance between robustness and imperceptibility perfectly. So, a zero-watermarking scheme is proposed to solve the problem. The proposed scheme can solve the contradiction between robustness and imperceptibility perfectly without introducing any audio quality degradation. The main processes are as follows: Firstly, constructing a binary pattern through the extracted audio features. Then, an exclusive OR (XOR) operation is performed between the pattern and the watermark image to get a key. Finally, the key is saved for watermark detecting. The proposed scheme uses the characteristic of the steady sign of certain DWT-DCT (DCT performed on DWT coefficients) coefficients with maximal absolute value to construct the binary pattern. The theoretical analysis and simulation results prove the effectiveness of the proposed scheme.