Novel Multi-bit and Multi-image Steganography Using Adaptive Embedding Algorithms with Minimum Error

This paper presents a novel multi-bit and multi-image steganography system using adaptive embedding algorithms with minimum error. These embedding algorithms can embed multi-bit (k-bit, k≥1) secret data into cover data only introduce minimum embedding errors which are restricted to ±(2
^k-1
-1)2
^i-k
according to the embedding location i. To achieve the goal of minimum embedding error, we introduced the adaptive embedding algorithms to adaptively evaluate the most similar value to replace the original one and which can be divided into three steps: (1) embed logo data into cover data, (2) adaptively adjust the least-significant bits (LSBs) of cover data, and (3) adaptively adjust the maximum-significant bits (MSBs) of cover data. Thus, we called them as adaptive 3-step embedding algorithms. The proposed embedding algorithms are not only achieving minimum error but also suitable to hardware implementation due to they are bit wise and based on logic, arithmetic and bit operations. After development of the adaptive 3-step embedding algorithms, we then built the multi-image steganography system to complicate the experiments of steganography. Many simulations show that the proposed adaptive 3-step embedding algorithms perform good embedding quality for steganography applications.