Steganalysis of Syntactic Transformation based Steganography

Syntactic transformation based(ST) steganography performs embedding operations of secret information by meaning-preserving syntactic transformations into sentences of a text. The syntactic transformations caused the difference of the occurrence frequency among different semantically equivalent syntactic structures vary between the cover texts and stego texts. A steganalysis method is therefore proposed using statistical features from these differences to detect the presence of secret information embedded by ST steganography. To verify the effectiveness of the proposed method, experiments were conducted on texts of two different genres in English. The texts are parsed by using the Stanford parser to extract typed dependency parses of sentences, and then the syntactic structures potential employed by ST steganography are automatically identified to obtain the discriminating features. The experimental results show that the proposed method provides high detection accuracy for various embedding rates and texts of different genres.