Resumen
Steganalysis is used for preventing the illegal use of steganography to ensure the security of network communication through detecting whether or not secret information is hidden in the carrier. This paper presents an approach to detect the quantization index modulation (QIM) of steganography in G.723.1 based on the analysis of the probability of occurrence of index values before and after steganography and studying the influence of adjacent index values in voice over internet protocol (VoIP). According to the change of index value distribution characteristics, this approach extracts the distribution probability matrix and the transition probability matrix as feature vectors, and uses principal component analysis (PCA) to reduce the dimensionality. Through a large amount of sample training, the support vector machine (SVM) is designed as a classifier to detect the QIM steganography. The speech samples with different embedding rates and different durations were tested to verify their impact on the accuracy of the steganalysis. The experimental results show that the proposed approach improves the accuracy and reliability of the steganalysis.