A novel detection scheme for MP3Stego with low payload

Abstract

MP3Stego is a typical steganographic tool of MP3 audios. Though many researchers have been making every effort on attacking it, the performance of their approaches could be improved especially at the low embedding rate. In this paper, we have proposed a scheme for detecting low embedding rate of MP3Stego. Based on investigating the embedding principle of MP3Stego and observing the alteration of quantized MDCT coefficients (QMDCTs), the one-step transition probabilities of the difference of quantized MDCT coefficients were extracted. Finally, SVM was used for constructing a classification model according to the extracted features. Experimental results show that our scheme can effectively detect the MP3Stego steganography with low payload.