REFERENCE-FREE STEGANALYSIS VIA NOISE ESTIMATION IN THE HISTOGRAM-CORRELATIVE DOMAIN

Abstract

It is difficult to ensure a reliable database of original images for the purpose of classification to detect image tampering or steganography. Hence reference-free analysis is necessary in care no database is available. In this paper we present a reference-free technique for detecting image steganography in the spatial domain. No need for an image database. The approach is based on estimating the noise introduced in the histogram autocorrelation of the cover image by LSB tampering while hiding a secret message in the cover image. Simulation results have shown the effectiveness of the proposed technique.