Research On Audio Information Hiding Algorithm Based On Neural Network

With the development of network communication and information processing technology, information hiding as a new information security technology is more and more widely researched and implemented in the aspect of covert communication. According to information hiding technology, specific information is embedded in kinds of digital media like image, audio, video, text and so on. Thus the purpose of information hiding is achieved. Meanwhile, the effect due to the specific information is not enough to attract people's attention. The information could be restored by the specific recovery method. And the information is invisible and not easily aware for illegal receiver.At present, the researches of information hiding mainly focus on the audio digital watermark. The watermark is a binary image. And the amount of hiding information is very small, and the state on one coefficient of hiding data is only 0 or 1. The quantization technology is adopted by the algorithm of embedding blind watermark. Then the blind detecting is realized by embedding the information into the different intervals.In order to change this situation, a new hiding algorithm based on neural network theory is studied in the thesis. Take the embedded grey image as the study object. And Perceptron neural network, BP neural network and Elman neural network are respectively used to hide and detect the grey image hided in the audio information. Under the situation of different train points and embedding strength, the theories and technologies which are suitable for hiding grey image will be searched for. And the great amount of data could be hided by the algorithm. The blind detecting of grey image could be realized. The anti-attacking robustness is strong. The speed of extracting hiding information is fast.The main research contents of this dissertation are recapitulated as follows:(1) The development situation and research focus of the information hiding and neural network at home and aboard are introduced. The characteristic and the application of the information hiding theory are introduced. The characteristic and deficiency of current information hiding technology are analyzed. And the information hiding algorithm based on neural network is researched. The research and application of neural network in the information hiding is mainly expatiated.(2) The fundamental conception, general model and learning rules of neural network are researched. And the fundamental characteristics of Perceptron neural network, BP neural network of feed forward neural network and the Elman neural network of recurrent neural network are introduced respectively. The researches on the algorithm of information hiding based on these neural networks are carried out.(3) According to BP neural network and Elman neural network, the effect of hiding and extracting image due to different training points and embedding strength is researched. According to the ability of attacking audio by Gaussian noise, a large number of simulation experiments are carried out. The experimental results and the analysis have been given out.It is indicated in experimental results that the algorithm of audio information hiding based on neural network has a lot of advantages. A great deal of data can be hided by the algorithm. And the time of extracting data is short and the anti-attacking robustness is strong.