Study On The Technique Of Digital Audio Color Watermarking And Multi-Watermarking

At present, digital multimedia products are widely spread on the internet, information transmission and sharing become more and more conveniently. But the problem of non-authorized copying appears consequently. Implementing copyright protection and rights protection on the internet has been one of the most urgent problems in the field of digital communication. The rise and development of digital watermark technique becomes a novel solution to handle this problem. With the gradual development of digital audio watermark technique, the musical products, represented by mp3, have been widely transmitted on the internet, it is of great theoretical and realistic significance to study digital audio watermark technology.The digital audio color watermarking and multi-watermarking technology is mainly studied in this paper. Three digital audio watermarking schemes are presented.(1) An adaptive quantification color audio watermarking scheme based on BTC. Considering the large embedding capacity of color watermarking image and the bad robustness of the traditional quantization embedding methods, using the compression characteristics that BTC coding did to the watermark image, combining the audio signal's characteristic with human auditory masking characteristic, the watermarking information is embedded in the low-frequency coefficients of the original audio in a adaptively quantized way, thus the colorful image watermark is quickly embedded and extracted, so the goal of copyright protection is achieved.(2) A wavelet-domain multi-purpose zero-watermarking method based on the characteristic of speech signal. By judging the detail component, the fragile watermark sequence is generated without any outside watermark data. The robust watermark image is selectively embedded in the key according to some important characteristics of the approximation component of the original audio signal, such as energy,4-order cumulant, etc. The ability to withstand the attacks is increased.(3) An adaptive quantification multi-purpose audio watermarking scheme based on LU decomposition. According to the 4-order statistics of the signal and the norm, the quantification step length is adaptively determined. The robust watermark information is quantitatively embedded in the U component which is obtained through the LU decomposition. The fragile watermark is embedded through the linear instantaneous mixing model. By using the FastICA to extract the watermark, the method shows a comparatively fragile sensitivity.Experiments show that the given three watermarking algorithms have strong safety and good transparency. Without the original audio signal, the watermark extraction can be extracted. Also, these methods have the ability to withstand the common signal processing and attacks. The robustness can be achieved through these methods.