Research And Realization Of Robust Digital Audio Watermarking Algorithm

With the development of digital multimedia and networks technology, multimedia digital work is easier to be composed, storaged and transmitted. This especially can be embodied in the wide spread of MP3 music in the Internet. However, unscrupulously duplicating and spreading pirate music work or distorting the content with copyright make the benefit of the work owners and issuers damaged greatly. Therefore, the copyright protection and information security in the network become more and more important. Traditional encryption techniques, which can only provide limited extent of protections, are insufficient in security and in communication. Digital watermarking technology, considered as a potential solution, has brought the cause to the attention of the world and is becoming an international academic focus.This dissertation focuses on research on watermarking algorithms of digital audio. The main content includes:(1)Through inducting fuzzy clustering theory, a new blind digital audio watermarking algorithm based on Discrete Wavelet Transform(DWT) is presented in this paper. The watermark can be embedded into the low frequency components of DWT domain by adaptive quantization according to the local audio correlation. The scheme effectively overcomes the lack of uniform quantization strategy.(2) On the basis of Support Vector Regression(SVR) theoretic and synchronization code technique, a novel digital audio watermarking scheme is proposed. The algorithm can select the stable energy and zero crossing rate of the audio as the feature vector and attain the SVR training model, which can be used to realize adaptive embedding position. By adopting quantization modulating strategy, the synchronization code and watermark are embedded respectively into the DCT and DWT domain to raise the resisting ability against various attack.