Research Of Adaptive Audio Watermarking Algorithm Based On Content

With the development of the digital multimedia technology, download of theaudio from internet generates heavy overuse for the audio resources. And the spreadof unauthorized copies seriously violates the intellectual property of related producers.The audio watermarking technology is widely applied in fighting against piracy andprotecting intellectual property as its unique advantages. Simultaneously, it isincreasingly attracted particular attention of researchers in the relevant fields.The adaptive audio watermarking algorithm base on content is proposed tocharacterize audio signal itself the mathematical characteristics of content as thereference factors of watermark embedding. If the change of the characteristics ofaudio signal itself’ content is more than a certain limit after normal signal processingoperation or malicious attacks, the change, audio signal will be treated as damaged,and lost its practical value. In this paper, the two designed algorithms, are based onaudio signal their own characteristics, through the corresponding mathematicalprocessing, extract the stable feature points of audio signal as a watermark embeddingposition reference, the applicable scope of the algorithm are more widely, theembedded watermark can not only guarantee the imperceptibility but also show strongrobustness on the common signal processing and synchronization attraction et al.In order to enhance the ability to stand against the anti-synchronization attractionof audio watermarking algorithm, an adaptive audio watermarking algorithm based onsub-band feature is proposed. Firstly, the Logistic mapping is used to scramblingprocess the watermark image, and the watermark information is encrypted by the chaosaddress index sequence. Secondly, the audio frame is divided into sub-bands tocalculate the spectrum centroid. Thirdly, the sub-bands which include the centroid, havethe small short-time zero-crossing rate, and have the large short-time energy chosen.Then, the chosen sub-bands are three lifting wavelet transformed to discrete cosinetransform the obtained low frequency coefficient. At last, the watermarking isembedded by the process of odd-even quantization. The original audio signal is not usedwhen watermark signal is detected or extracted. Experiments show that thewatermarking not only has good concealment, but also has strong robustness afterlow-pass filtering, re-sampling, format conversion, synchronization attacks and otheroperations.Aiming to solve the audio quality problems which is subject to be varied attacked and disturbed to get worse quality of watermarking information extracted in thecourse of audio transmission, a new robustness audio watermarking algorithm basedon singular value decomposition (SVD) is presented. Firstly, the Baker transform isused to scrambling processing the watermark image, and the watermark information isencrypted by Logistic mapping for the images in disorder. Secondly, the audio signalis divided into appropriate length audio frame. From the standpoint of matrix, eachaudio frame is transformed from one-dimensional to two-dimensional matrix. And doa singular value decomposition to the two dimensional matrix; Then using the stabilityof the gained characteristic value, combined with the watermarking embeddingcoefficients, complete the embedding of watermarking information; Finally, accordingto a certain order, transform the gained two dimensional matrix into one dimensionalmatrix, then according to the order of the beginning division, connected the audioframe, now we get audio signal contained watermarks. The simulation results showthat the algorithm could preferably control watermarking embedding coefficients andensure the imperceptibility, meanwhile show good robustness in the face of the normalsignal processing operation and a certain degree of malicious attacks.