Design Of Digital Audio Watermarking Method Based On Dual-tree Complex Wavelet Transform

With the continuous progress of modern Internet technology,the convenience and applicability of information preservation have been greatly improved,which provides a lot of convenience for the masses,but it has also formed many aspects of influence.For example,copyright infringement issues continue to occur.Especially in recent years,the frequent illegal editing and embezzlement of original video,these problems are very unfair for the original author.For the audio digital watermarking technology,its focus is the secret watermark data in the audio and video(including image,video and so on),so as to test the legitimacy of audio.This technology has been applied in many fields,including copyright protection,secret communication and so on,which makes it become a hot research topic of relevant scholars.This paper mainly analyzes the related problems of digital watermarking based on audio,deeply studies the watermarking algorithm in transform domain,and designs and implements two watermarking algorithms based on transform domain for robust watermarking.The main contents of this paper are described as follows:(1)In view of the imbalance between robustness and imperceptibility,a watermark algorithm based on discrete wavelet transform and singular value decomposition is designed.After segmenting the original speech,the three-level wavelet transform is completed for each segment,then the low-frequency component is selected to construct the matrix,the singular value is solved,and the watermark embedding can be realized by adjusting the singular value.The method can take into account the robustness and the imperceptibility by selecting the suitable embedding strength.The experimental results show that the accuracy of watermark extraction reaches more than 97%.(2)As the watermark content increases,the robustness and imperceptibility of audio will decrease.An audio watermarking algorithm based on dual tree complex wavelet transform is designed.In the embedding stage,the carrier audio is decomposed into three levels by dual tree complex wavelet transform.The low-frequency information is selected,and then the singular value decomposition is completed.The watermark is embedded effectively by adjusting the singular value.The final results show that,compared with the algorithm based on wavelet transform,this algorithm can not only ensure the good robustness of the carrier audio,but also can hide and extract the image secret watermark.The experimental results show that compared with the watermarking algorithm based on wavelet transform,the signal-to-noise ratio is increased by 10%,and the watermark capacity is increased by more than one time.