An improved reversible data hiding scheme using best neighboring coding based on color images

Abstract

In this paper, an improved reversible data hiding (RDH) scheme, best neighboring coding (BNC), is proposed for vector quantization (VQ) compressed color images. As a result, it improves joint neighboring coding (JNC) based schemes, which take advantage of correlation between indices to embed secret bits (SB) and achieves good compression rate. As the correlation between RGB channels and the information of reference channel are employed, the best neighboring blocks of the other channels are obtained. The experimental results indicate that BNC enhances the prediction possibility of the best neighboring index by about 2.5 times compared to the conventional JNC method, which applies the current index and one of its four neighbor indices pointed by the secret data to generate the difference. Consequently, a more concentrated distribution of difference between current index and its neighboring index is obtained. After encoding this distribution generated by RGB channels, better compression rate and higher embedding efficiency are derived. Furthermore, extended experiments demonstrate that the proposed method outperforms other state-of-the-art RDH methods as well.