Steganography preserving statistical properties

Abstract

Steganographic modifications must not noticeably change characteristics of the cover data known to an attacker. However, it is a difficult task to define an appropriate set of features that should be preserved. This article investigates possibilities to maintain statistical properties while embedding information. As an example, the simple embedding function "overwriting the least significant bits" is discussed, w.r.t. grayscale images used as covers. Two modifications of this technique are presented here: The first modification avoids changes of first-order statistics and thereby overcomes possible attacks based on histogram analysis. The second modification tries to reduce noticeable modifications of second-order statistics and, thereby, of the structure of an image. The suggested modifications are tested on a set of images. Testing a possible attack and comparing these results to the results of the same attack on stego images created by the original method clarifies the improvements. © Springer-Verlag Berlin Heidelberg 2003.