Digital Audio Watermarking Based On Prediction Error Expansion And SVD

While the internet provides an easy channel for digital audio resource,it also brings forward new issues on intellectual property protection.The situation of copy-right violation,various attacks through the internet have been very serious.Audio watermarking entered people's vision.In recent years,audio watermarking has attacked a lot of attention with important applications in audio copyright protection,secure communications and so on.We classified the audio watermarking algorithms into two categories according to their characteristics.One is robust audio watermarking and the other is fragile audio watermarking.Singular value decomposition(SVD)is an important signal decomposition tools,it has the excellent characteristic of stability which meets the robust requirements of watermarking algorithm.Most watermarking algorithms based on SVD have a good robustness and imperceptibility,but they usually have the drawback with the payload.Prediction error expansion can be used to design reversible watermarking algorithm.It has been widely used in image application but less in audio.Currently,audio reversible watermarking algorithms have the problem with the poor imperceptibility and the large auxiliary information.This article launches the research based on the above problems.The specific work is as follows:We proposed a novel blind audio watermarking algorithm,which combined SVD with Discrete Wavelet Transform(DWT).In our algorithm,we first partitioned the rearranged audio signal into blocks,and then generated the vector by selecting the biggest singular values after performing SVD on these blocks.Finally we embedded the watermark into the approximate components obtained from the DWT decomposition of the vector by means of quantization process.Experimental results showed that our robust watermarking algorithm has good robustness against the common audio signals processing operations.Compared with earlier schemes based on SVD,the proposed scheme has satisfying imperceptibility and improved payload.We presented a novel reversible audio watermarking scheme based on improved prediction error expansion.First,we used differential evolution algorithm to optimize prediction coefficients,and then applied prediction error expansion to output stego data.Second,in order to reduce location map bits length,we introduced histogram shifting scheme.Meanwhile,the corresponding prediction error modification of a given embedding capacity can be computed by our proposed scheme.Experiments showed that this algorithm improved the Signal Noise Radio(SNR)of embedded audio signals and embedding capacity,drastically reduced location map bits length and enhanced capacity control capability.