VQ image steganographic method with high embedding capacity using multi-way search approach

Abstract

Embedding large quantities of data in VQ (Vector Quantization) images is a thorny problem, since the hiding schemes usually have to change the index values of the VQ images, which might cause serious image distortion. As a result, many currently existing methods can only afford to support a small embedding capacity. In this article, we shall propose a new method that uses the genetic clustering technique on the codebook to obtain better clusters so that the replacement distortion of indices can be reduced. Then, we apply multi-way search to hide the secret data. Experimental results show that our new method outperforms existing schemes on both image quality and embedding capacity. © Springer-Verlag Berlin Heidelberg 2005.