Research On Robust Compressed Audio Watermarking

With the development of the Internet and mature of audio compression techniques, the network music, such as MP3, is widely applied to Digital Audio Broadcasting, Audio-on-Demand and Online Music Sales Systems. Therefore, the problems of copyright protection and security authentication for audio media become increasingly prominent. As an effective complement of traditional Encryption, digital watermarking is a new technology to protect copyright and authenticate source or integrality of the media in an open network environment, and in recent years it has become one of the research focuses in the information security domain. The thesis mostly combines with MP3 coding Standard, and makes some research on robust compressed audio watermarking techniques.The thesis first gives a comprehensive summarization about the digital audio watermarking, then integrates the idea of localized watermarking technique and invariant watermark, and proposes a novel combined audio compression/watermarking algorithm based on Localized-Invariant-Feature (LIF) by analyzing the synchronization problem which exists in traditional algorithms. In order to bring down the Bit Error Rate (BER), the paper also proposes a robust compressed audio watermarking algorithm based on concatenation of BCH codes and repetition codes. The main works are as follows:1. The related knowledge of audio watermarking technique is first introduced; then the typical watermarking algorithms for both uncompressed and compressed audio are introduced and analyzed; finally, the main attacks against current digital watermarking with their solutions are summarized, and the evaluation framework for robust audio watermarking is also discussed.2. A LIF-based combined compression/watermarking algorithm for audio signals is presented. The localized embedding regions that corresponding to music edges like audio rhythms and beats, are first selected by adopting an audio content analysis in the audio compression procedure, then the Invariant Statistic Feature (ISF) that is insensitive to signal processing and synchronization attacks is extracted for each regions. Finally, the watermark is modulated by a random chaotic sequence and embedded into the ISF of localized regions. The experimental results demonstrate the algorithm is quite imperceptible, and the robustness against common audio signal processing and synchronization attacks, such as random cropping, jittering and TSM (Time Scale Modification), is also improved comparing to traditional compressed audio watermarking algorithms.3. Aiming at the high BER cases existing in the above algorithm under some attacks, a robust compressed audio watermarking algorithm based on concatenation coding is presented. The watermark is coded by concatenation of BCH codes and repetition codes to make the abstracted watermark sequence having correction capability, so that the watermark's BER can be brought down further. Furthermore, according to the analysis from the experiments' result, several conclusions about the application of concatenation codes in audio watermarking algorithm are obtained in the end.