Generalizing Hamming+k data hiding by overlapped pixels

Abstract

Matrix coding based data hiding (MCDH) using linear codes (syndrome coding) is an efficient coding method for steganographic schemes to improve their embedding efficiency. Hamming code data hiding (HDH) is a well-known MCDH using a covering function COV (1, n = 2k − 1, k). Afterwards, Hamming+ 1 DH (H1DH) was proposed with good embedding efficiency. However, these two previous approaches, HDH and H1DH, are not efficient for a large amount of messages. To tackle this problem, Yang et al.’s proposed Hamming+ k DH (Hk DH), which enhance the extra one embedded bit in H1DH to k embedded bits in the Hk DH. In this paper, we extended the Hk DH to the Hamming+ k with m overlapped pixels (Hk_mDH). The proposed Hk_mDH adopted pixel overlapping approach, optimal pixel adjustment process (OPAP), and Least Significant Bit (LSB) substitution. Experimental results demonstrate that our Hk_mDH has better embedding rate (ER) compared with previous schemes. In addition, we have proved that our Hk_mDH has excellent theoretical estimation of average mean square error.