Novel multi-bit and bitwise minimum error replacement technique for watermarking and data hiding

This paper presents an novel bitwise minimum error replacement (MER) technique for data hiding. Such a bitwise MER can embed k-bit (k ≥ 1) secret data into cover data only introduce minimum embedding error smaller than (2^k-1 - 1)2^i-k according to the embedding location i. To achieve the goal of MER, we develop novel multi-bit bitwise 3-step MER algorithms to adaptively evaluate the most similar value to replace the original one, which consist of three steps: (1) embed logo data into cover data, (2) adjust the least-significant bits (LSBs) of cover data, and (3) adjust the maximum-significant bits (MSBs) of cover data. Reason that we called these multi-bit MER algorithms are "bitwise" is that data replacements are based on the logic and algebraic bit operations to achieve minimum error. Many simulations show that the proposed MER performs good embedding quality for watermarking and steganography applications.