The Research On Information Hiding In Compressed Audio

With the rapid development of internet and multimedia technology, multimedia communication has become one of the most important means to exchange information, and information security and protection is becoming increasing important. Information hiding is considered one of the most effective measures to solve multimedia security issue. According to the different audio carrier format, audio information hiding is divided into two categories: data hiding in compressed domain and data hiding in uncompressed domain. In the application fieled, audio signal is aways in compressed format for transmission and storage in network, so the compressed domain information hiding is more pratical significans than uncompressed. In this paper, we study on MP3 audio. The main work is concluded as following several aspects.1. Make an in-depth analysis of the MP3 encoding standard theory, which focuses on sub-band filter banks, MDCT, psychoacoustic model, quantization and coding, bit stream format.2. Summarize available compressed audio information hiding method, and we divide it into three categories which are hiding method base on time-frequency conversion, hiding method base on quantization and coding, hiding method base on bit-stream.3. Propose two hiding methods base on quantization and coding. The first method makes use of the noise controller in quantization process, and conbine with psycho-acoustic model, the secret message will be inserted iteratively until without distortion. This method can obtain the high capacity and good imperceptibility. The second method is to consider the security point of view, insert data by modifying quantization step. This method is similar with MP3Stego but superior to it.4. Propose a hiding method base on bit-stream, this method insert data by swapping two VLC codes which with the same size of code word and the same size of single bit. To further increase the embedding capacity, we also proposed two improved methods. The first makes code words to be paired with each other, so as to increase the replaceable code. The second is to swap with more than two code words, so that each swapping can embed more than one bit, thereby greatly increase the embedding capacity.