A further study of large payloads matrix embedding

• A further study of large payloads ME by improving the embedding efficiency is proposed.
• A new matrix is constructed for ME.
• A sub-optimal search strategy is exploited to reduce the computational complexity.
• Effective and practically feasible large payloads ME can be realized by the proposed method.

Matrix embedding (ME) is a well-known steganographic method that can improve the embedding efficiency. In ME, the sender and recipient agree in advance on a parity check matrix (PCM) of a binary linear code and utilize the PCM for data embedding. At the decoder side, the embedded message can be extracted by the recipient as the syndrome of the received stego data. The PCM can be taken as any full-ranked matrix and its selection is actually crucial to the embedding performance. In this paper, by extending some previous works, we propose a novel scheme to further improve the embedding efficiency of large payloads ME. First, we utilize a new and specifically designed matrix for ME. Then, instead of finding a coset leader as the modification to the cover which is time consuming, we turn to finding a vector in the coset with relatively small Hamming weight. By the proposed approach, effective and practically feasible large payloads ME can be realized, and extensive experiments show that, a significant increase of embedding efficiency is achieved compared with some state-of-the-art works.