Stego anomaly detection in images exploiting the curvelet higher order statistics using evolutionary support vector machine

Steganalysis is an important extension to existing security infrastructure, and is gaining more research focus of forensic investigators and information security researchers. This paper reports the design principles and evaluation results of a new experimental blind image steganalysing system. This work approaches the steganalysis task as a pattern classification problem. The detection accuracy of the steganalyser depends on the selection of low-dimensional informative features. We investigate this problem as a three step process and propose a novel steganalyser with the following implications: a) Selection of the Curvelet sub-band image representation that offers better discrimination ability for detecting stego anomalies in images, than other conventional wavelet transforms. b) Exploiting the empirical moments of the transformation as effective steganalytic features c) Realizing the system using an efficient classifier, evolutionary-Support Vector Machine (SVM) model that provides promising classification rate. An extensive empirical evaluation on a database containing 5600 clean and stego images shows that the proposed scheme is a state-of-the-art steganalyser that outperforms other previous steganalytic methods.