Research On Robust Audio Watermarking Algorithms

In recent years, digital product copyright issues have become increasing prominence so thatdigital watermarking technology has attracted the attention of many experts and scholars in thiscase.The thesis has researched mainly on the algorithms of audio digital watermarking. Itanalyzes the theory of audio watermarking and the characters of audio and implementations ofcommon watermarking algorithms. From the point of improving the watermarkingimperceptibility and robust, three digital audio watermarking algorithms are proposed based onthe analysis and comparison of existing algorithms. The main research work are as follows:(1) An gray image embedded audio watermarking algorithm is proposed based on BTC.Firstly a gray image is compressed to reduce the impact on the audio carrier. Then the audio issub-framed adaptively and averaged for each sub-frame so as to resist to scaling attacks.Secondly the binary image obtained by BTC is embedded into the low-frequency coefficients ofthe transform domain of audio carrier, with the purpose of resisting many common attacks andcopyrighting authentication in high visual effects.(2) An audio zero-watermarking algorithm is proposed based on the mean valueinterpolation operation. Firstly the approximate component of three level wavelet decompositionof the original audio, then it be processed by discrete cosine transform. The DCT coefficients ofeven-numbered points in the selected number are applied mean interpolation operation togenerate a new coefficient set, producing audio characteristic sequence with comparison with theoriginal even point coefficient. Finally the characteristic sequence can do XOR with thecopyright watermark image to get copyright information.The simulation results show that thealgorithm has good robustness characteristic.(3) Considersing that zero-watermarking algorithm in chapter four is not in good conditionon resisting to partical attacks, an robust audio zero-watermarking algorithm in DCT domain isimproved. Audio characteristic sequence can be structured by the DC component set, because itis not easy to be changed. Applying the BCH coding technique to image of chaotic encryptioncan reduce the bit error rate under attack. The simulation results show that such a zero-watermark algorithm has very good robustness characteristic.