Title of Paper:Steganalysis of SS Steganography: Hidden Data Identification and Extraction

Hits:

Date of Publication:2015-10-01

Journal:CIRCUITS SYSTEMS AND SIGNAL PROCESSING

Included Journals:SCIE、EI、Scopus

Volume:34

Issue:10

Page Number:3305-3324

ISSN No.:0278-081X

Key Words:Blind detection; Covert communications; Data hiding; Spread-spectrum embedding; Steganalysis; Steganography

Abstract:We consider the problem of steganalysis of multi-carrier spread-spectrum steganography in which secret data are embedded over a wide band in a spectrum (transform) domain of a digital medium. The objective is first to identify the presence or absence of hidden data in a given image (i.e., passive steganalysis). Unlike conventional feature-based passive steganalysis algorithms, we describe an unsupervised (blind) low-complexity approach based on iterative generalized least-squares principles that may enable rapid high-volume image processing. Extensive experiments on image sets and comparisons with existing passive steganalysis techniques demonstrate most satisfactory classification performance measured in probability of correct detection versus induced false alarm rate. If the existence of hidden data is identified, we then aim to detect the number of carriers (messages) and blindly extract the hidden data without the knowledge of the original host nor the embedding carriers. This task is also known as active/forensic steganalysis. Experimental studies on images show that the developed algorithm can achieve recovery probability of error close to what may be attained with known embedding carriers.