Adaptive data hiding for vector quantization images based on overlapping codeword clustering

In this paper, an overlapping codeword clustering based data hiding scheme is presented. In this scheme, a mapping table is designed to determine the overlapping codeword clustering and to indicate the index modification in the secret embedding. The mapping table explores the relationship among the sub-codebook’s size, the codeword’s order and the embedding secret message to which the codeword overlapping in sub-codebooks with different sizes is permitted. In addition, the secret embedding is also determined according to the mapping table. The experimental results showed that the number of partitioned sub-codebooks was increased significantly. The average hiding capacity was about 30 K bits while the average embedding distortion was about 1.2 dB. In comparison to similar methods, the proposed scheme provided a larger hiding capacity than others while preserving a similar stego-image quality. Furthermore, the proposed scheme offered a better proportion of hiding compared to image distortion.