An image based undetectable steganographic technique (ImUST)

Abstract

Steganography is a science of secure communication. It is challenged by a binary steganalyzer which predicts presence or absence of a hidden message within a cover media. Recently proposed quantitative steganalyzers go a step beyond binary detection and estimates the hidden message length. In this paper, we propose a novel image based steganographic technique. A preprocessing step is introduced to minimize distortion and the message is embedded using histogram shifting. The statistical features of an image do not change significantly after embedding. We show that the proposed technique remains undetectable to state-of-the-art; binary and quantitative steganalyzers while the message remains undetectable. The preprocessing also enables iterative embedding, generating higher steganographic capacity than state-of-the-art.