A Lossless Data Hiding Scheme Based on VQ Index Frequency

Abstract

With the rapid development of the digital multimedia and network communication, data hiding in the compression domain becomes increasingly popular in the need for speeding up the transmission process and reducing bandwidth. Recently, various kinds of watermarking techniques based on VQ have attracted more interests from researchers. In 2015, Tu and Wang proposed a reversible data hiding scheme with high payload which embeds confidential information into the VQ-compression codes of a host image with a relatively higher capacity than other existing methods. However, their scheme introduced extra bits to reverse the original vector quantization index of the cover image. In this study, we propose a novel reversible hiding scheme which reduces the usage of extra bits. In our method, a codebook is clustered into six groups which are organized in accordance with the referred frequency of codewords from high to low. Then, a new mapping relationship among the six clusters is constructed to hide secret data more cost-effectively with high payload. The experimental results demonstrate that the proposed scheme outperforms Tu and Wang's scheme for complex texture images.