Contemporary stegnalysis schemes for reliable detection of steganography

Abstract

Steganalysis is the process of detecting the hidden information in the carrier. Most used carriers for steganography are images due to the redundant information present in the images and frequency of their use on the Internet. Steganalysis methods are classified into two categories, Targeted steganalysis and universal steganalysis. Targeted steganalysis is based on analysis of individual and known steganographic scheme. Blind steganalysis methods detect steganographic schemes created by unknown random stego-systems. The objective of steganalysis algorithms is to distinguish stego images from pure images. A classifier is built based on stego and pure images. When the knowledge of steganographic scheme is not available, a general steganalyzer is built, which is trained with a set of pure images and a set of stego images generated by various steganographic algorithms. The performance of steganalysis algorithm depends on three important aspects, preprocessing technique, feature selection & extraction and classification. This paper presents the contemporary steganalysis schemes discussing the details and comparing various aspects of these methods.