Multi-bit Adaptive Embedding Algorithm for Anti-forensic Steganography

Multi-bit minimum error replacement (MER) is a method that can embed multi-bit logo/secret data into k least-significant bits (LSBs) of cover data only introduces minimum embedding error (MEE). However, k-LSBs MER suffers from weak anti-forensics. Moreover, it is unfortunate because other previous steganography works have seldom considered both large embedding capacity and high image quality. Therefore, this work proposes an anti-forensic steganography system using multi-bit adaptive embedding algorithm with flexible bit location to overcome the problem of forensics and to achieve high performance includes both large embedding capacity and high image quality. The proposed embedding algorithm embeds and hides multi-bit (k-bit, k ≥ =1) logo/secret data into any adjoining (starting with the i^th location) k-bit b_ib_i-1⋯Kb_i-k+1 of cover data only introduces MEE in the range of 0 to -±(2^k-1)2^i-k, moreover, its flexible bit location enhances the embedded security as security increases as embedding location increases. As the proposed embedding algorithm can embed multi-bit logo/secret data into any adjoining bits of cover data and has large embedding capacity and high embedding quality, this method was applied to develop image steganography systems. Finally, anti-forensics of the proposed steganography systems are demonstrated using the visual attack and the statistical attack of Chi-square analysis.