Research On Digital Audio Watermarking Technology Based On Compressed Sensing

With the advent of the era of big data,any files like text,audio and video can be widely spread on the Internet.While enjoying the convenience it takes,multimedia information security and copyright issues followed.If these problems can not be solved very well which will lead to the unhealthy development of the entire electronic publishing industry(including books,music and movies).The digital audio watermarking technology can hide the copyright information in the original audio carrier through the corresponding algorithm,not only can encrypt the original carrier,but also protect the copyright.Therefore,it is of great practical significance to research the copyright protection technology of digital product.In this thesis,two audio watermarking algorithms are proposed for the copyright protection of audio signals.The audio watermarking system is analyzed and researched in terms of security,imperceptibility and robustness.Firstly,using the theory of compressed sensing to project the binary image on the orthogonal Fourier and orthogonal cosine transform bases respectively,and based on the experimental results,the orthogonal cosine transform base with better sparse effect is selected as the sparse basis of the watermark image.Secondly,according to the restricted-isometry property(RIP),a random Gaussian matrix is selected as the observation matrix.Finally,the classical orthogonal matching pursuit algorithm(OMP)is used to reconstruct the watermark image.According to the experimental results,the watermark extraction effect is enhanced with the watermark embedding capacity.Then,a digital audio watermarking algorithm based on compressed sensing and discrete wavelet transform(DWT)is proposed.The algorithm utilizes the characteristics that the energy of the audio carrier signal is mainly concentrated in the approximate component of the low frequency in the discrete wavelet transform domain,and the modification of the larger coefficient can improve the transparency of the watermark.Firstly,the Arnold scrambling algorithm is used to encrypt the watermark.Then,using the compressed sensing theory,the scrambled watermark information is projected on the orthogonal cosine transform base,and then the random Gauss matrix is selected as the observation matrix.The watermark is embedded in the discrete wavelet transform coefficient of audio,and finally the watermark is extracted by the OMP algorithm without using the original host signal,and the Arnold inverse transform is used to restore the original watermark image.In order to further improve the security of the watermark algorithm,a digital audio watermarking algorithm based on chaotic encryption and DWT is proposed.The algorithm uses the randomness and initial value sensitivity of the logistic chaotic system to encrypt the watermark image,so that the original watermark cannot be correctly extracted without the encryption key.Then the discrete wavelet transform is performed on the audio carrier signal and the approximate components are taken to perform the discrete cosine transform(DCT).Finally,the encrypted watermark information is embedded into the transform coefficient of the audio carrier according to the embedding algorithm.The extraction process and the embedding process are mutually inverse processes.Finally,the simulation experiments and comparisons is carried out between the proposed two algorithms and a sparse watermarking algorithm.Through the objective analysis of the peak signal-to-noise ratio and the normalized similarity of the simulation results,the imperceptibility of the compressed sensing algorithm is better than the chaotic encryption algorithm,and its robustness of low-pass filtering,resampling and amplitude compression is better.The security of chaotic encryption algorithm and its anti-shear attack have certain advantages over compressed sensing algorithm and sparse watermarking algorithm.The two watermarking algorithms explored in the thesis both realize the blind extraction process,which provides a certain reference for the research of digital audio watermarking technology.