Research On Digital Audio Information Hiding Technology Based On Transform Domain

Audio information hiding technology is to embed imperceptible secret information into the audio signal to achieve the functions of copyright protection and covert communications, which includes two branches of audio watermarking and audio steganography. This dissertation aims to the performance requirements of digital audio information hiding, research several transform domain-based audio information hiding technology, the main work is as follows:(1) According to the human auditory model and the masking effect, an adaptive audio steganography of integer discrete wavelet packet domain was proposed. The method adopted integer lifting wavelet packet to reduce quantization error, and combined the auditory model in different bands with different masking threshold and the characteristics of wavelet coefficients to determine the embedding bits of the secret information into the audio median adaptively. The method has larger embedding capacity and higher imperceptibility.(2) Considering the characters of non-linear, input-output mapping, adaptive and fault-tolerance in neural network. A kind of robust audio watermarking method based on counter neural network was proposed. The ratio of transform coefficients and the corresponding values of the watermark were used to train the neural network. So the watermark was embedded into the neurons of neural network. The experimental results proved the method can resist some common attacks while keeping good robustness. And at the same time, the watermark embedding and extraction process gets simpler than the traditional method.(3) Considering the robustness and the inaudibility conflict each other in watermarking system, an adaptive audio watermarking method based on genetic algorithm was proposed. By defining the appropriate fitness function, using genetic algorithms to compromise contradiction between robustness and the imperceptivity in the watermarking system, the quantization index strength is adjusted and the audio segment for watermark is selected adaptively. Thus the optimal problem of robustness and imperceptivity in audio watermarking system is solved.In this paper, the research was verified and analyzed by Matlab7.0 simulation platform.