Research And Implementation On Robust Digital Audio Watermarking Key Technologies

With the development of the Internet and digital audio coding technology, transmission and access of digital audio are more and more facilities .At the same time the phenomenon of piracy makes the copyright protection of digital audio increasingly important. Digital audio watermarking technique is an effective means of achieving copyright protection. Combines wavelet analysis, human auditory masking, frame synchronization, as well as MP3 audio coding standards, this thesis designs a research framework for robust digital audio watermarking. According to the framework, it does some researches on the original audio and compressed audio watermarking technique, which includes the following three parts:Firstly, the thesis studies wavelet analysis and principles of human auditory masking. With the original audio as carrier, it proposes a watermarking algorithm based on wavelet packet transform and blindness of human auditory masking .In the embedding process, algorithm analyzes the characteristics of power spectrum in the wavelet packet field to calculate the threshold of auditory masking. According to the threshold, picks the signals which are difficult for the human auditory to perceive, and select the greater ones in energy as embedding positions. Algorithm significantly increases the imperceptibility, makes up the low accuracy in choosing embedding position, too subjective, which are shortcomings of traditional algorithm.Secondly, aims at the weakness of audio watermarking algorithm in resisting original anti-simultaneous attacks, the thesis quotes frame synchronization technique, proposes a blind watermarking algorithm for resisting anti-simultaneous attacks, which based on m sequences. In the algorithm, paragraph numbers as part of the watermark are embedded into the audio, while m sequence acts as frame synchronization codes is embedded into the audio in the wavelet packet field. The extraction process, by detecting the frame synchronization code, ensures that watermark paragraph numbers in accurate positioning for restructuring. Algorithm resist the time-domain samples cut, copy samples and other simultaneous attacks well.Thirdly, for the synchronization problem of the built-up compressed audio watermark, it proposes an algorithm based on local invariant features. Algorithm analyzes the content of compressed audio, selects the local area corresponding to the edge information as embedding position ,in each local area extracts constant statistical characteristics that not sensitive for the audio signal processing and synchronization attacks, and makes the watermark with chaotic modulation embedded into constant statistical features. Algorithm has better imperceptibility, effectively resist conventional audio signal processing and random shear, jitter, TSM and simultaneous attacks.