Reliable steganalysis of HUGO steganography based on partially known plaintext

Abstract

The image adaptive steganography represented by HUGO (Highly Undetectable steGO) has high anti-detection capabilities. The primary challenge for steganalyzers is how to reliably detect such steganography and extract the embedded message from stego. Although existing steganalysis algorithms based on parameter recognition of STCs (Syndrome-Trellis Codes) can reliably detect adaptive steganography when the embedded message is in plaintext, the steganalysis method is ineffective when the embedded message is in ciphertext. Therefore, a steganalysis algorithm based on partially known plaintext is proposed in this paper. The steganalysis algorithm targets situations wherein in order to facilitate the receiver’s extraction and storage of the embedded message, the file format name and message length may be transported without encryption when HUGO steganography is used to transport the encrypted file. First, the structural characteristics of the parity-check matrix were utilized to simplify the STCs decoding equation. Second, we calculated the submatrix by solving nonhomogeneous linear equations instead of exhausting submatrix, thus, parameter recognition efficiency was significantly improved. Finally, we verified the correctness of the submatrix and extracted the embedded message from stego images. The experimental results show that parameter recognition of STCs can be achieved using an ordinary PC within a short time. Therefore, the embedded message can be extracted when the partly embedded message is known.