Fuzzy based clustering for grayscale image steganalysis

Abstract EN

Steganography is the science that involves communicating secret message in a multimedia carrier.

On the other hand, steganalysis is the field dedicated to detect whether a given multimedia has hidden message in it.

The detection of hidden messages is revealed as a classification problem.

To this end, this paper has two contributions.

Up to the best of our knowledge, this is the first time todefinegrayscale image steganalysis, as a fuzzy c-means clustering (FCM) problem.

The objective of the formulated fuzzy problem is to construct two fuzzy clusters: cover-image and stego-image clusters.

The second contribution is to define a new detector, called calibrated Histogram Characteristic Function (HCF) with HaarWavelet(HCF??).

The proposed detector is exploited, by the fuzzy clustering algorithm, as a feature set parameter to define the boundaries of the cover- and stego- images clusters.

Performance evaluations of FCM with HCF?? in terms of accuracy, detection rate, and false positive rate are investigated and compared with other work based on HCF Center of Mass or HCF-COM andcalibrated HCF-COM by down sampling.

The comparison reveals out that the proposed FCM with (HCF??)significantly outperforms other work.