In this prospectus we are proposing to develop a unified Steganalyzer that can not only work with different media types such as images and audio, but further is capable of providing improved accuracy in stego detection through the use of multiple algorithms running in parallel. Our proposed system integrates different steganalysis techniques in a reliable Steganalyzer with distributed and Services Oriented Architecture (SOA). The distributed architecture not only allows for concurrent processing to speed up the system, but also provides higher reliability than reported in the existing literature. The extendable nature of the SOA implementation allows for easy addition of new Steganalysis algorithms to the system in terms of services. The universal steganalysis technique proposed in this prospectus involves two processes; feature extraction and feature classification. Three methods are used for feature extraction; Mel-Cepstrum and Markov (for audio), and Intra-blocks for (JPEG images). The feature classification process is implemented using neural network classifier. The unified steganalyzer is tested for JPEG images and WAV audio files. The accuracy of classification ranges from 96.8% to 99.8% depending on the object type and the feature extraction method. In particular, an enhancement of Mel-Cepstrum technique is proposed that achieves an accuracy of 99.8%. This is significantly better than detection accuracy of 89.9% to 98.6% [Liu 2011] where even a much larger training dataset was used than ours.

Citation: Poster 10