Strong Robustness Of The Digital Audio Information Hiding Technology And Safety Studies

With the rapid development and wide application of information technology,information delivery and communication is becoming increasingly simple and easy,but the information security problems follow. In the traditional field of informationsecurity, cryptology has been playing a vital role, but when facing the new applicationdemand and ever-increasing computing power, cryptology has been insufficient to meetthe requirements of information protection, thusly, the information hiding technologybegins to appear.Information hiding technology is to embed secret informations in digital images,videos, audios and other files in order to achieve covert communication and copyrightprotection and other functions. The key of information hiding is that it mainly coversthe covert communication events, but cryptology mainly covers the secret informations,this kind of difference happens to be a reasonable and effective complementaryrelationship, the combination of two methods can make better security of confidentialinformation. At present, the information hiding research is focused on the digital images,while the digital audio is widely disseminated, it is an ideal cover source, so we select itas the direction of information hiding research and do the following innovative work:1) Analysing the defects of the Least Significant Bit (LSB) algorithm, it takesnondistinctive embedding to cover audio segments without taking the characteristics ofsignal changes into account, which brings much noise, and the simple single samplingpoint embedding operation could easily lead to the law of the statistical histogram, it isdetrimental to resist the statistics detection. On this basis, we proposed to combine thefeatures of the audio signal segments with embedding, while introducing the short-termenergy and energy-zero-product concepts in speech recognition technology and usingthe grouped sampling points embedding method, which effectively improves theperformance of the algorithm.2) According to the energy characteristics of the audio signal, we proposed that using short-term energy judgement to dynamically adjust the echo intensity into eachsignal segment, and using forward and backward echo kernels to scatter the echo energy,it improves the imperceptibility of embedding. Instead of the cepstrum check, we usedcepstral autocorrelation method to detect echo, it improves the detection accuracy andenhances the robustness of the algorithm.3) According to the characteristics and advantages of the Discrete CosineTransform (DCT) and Discrete Wavelet Transform (DWT), combined with humanauditory perception features, we gave a mixed transform domain embedding algorithmbased on critical band. The algorithm utilizes the flexibility and locality of waveletdecomposition to divide the cover audio into approaching critical band, and embedssecret signals on DC component of DCT in appropriate frequency band. Theexperimental results show that the algorithm has good overall performance.