International Journal on Cryptography and Information Security (IJCIS),Vol.2, No.3, September 2012 DOI:10.5121/ijcis.2012.2307 77 Blind Image Steganalysis Based on Contourlet Transform Natarajan V 1 and R Anitha Department of Applied Mathematics and Computational Sciences, PSG College of Technology, Coimbatore, India. 1 kvn.psg@gmail.com 2 anitha_nadarajan@mail.psgtech.ac.in ABSTRACT This paper presents a new blind approach of image Steganalysis based on contourlet transform and non linear support vector machine. Properties of Contourlet transform are used to extract features of images, and non linear support vector machine is used to classify the stego and cover images. The important aspect of this paper is that, it uses the minimum number of features in the transform domain and gives a better accuracy than many of the existing stegananlysis methods. The efficiency of the proposed method is demonstrated through experimental results. Also its performance is compared with the state of the art wavelet based steganalyzer (WBS), Feature based steganalyzer (FBS) and Contourlet based steganalyzer (WBS). Finally, the results show that the proposed method is very efficient in terms of its detection accuracy and computational cost. KEYWORDS Steganalysis, Contourlet transform, Structural similarity measure, Non linear support vector machine 1. INTRODUCTION Steganography is the art and science of hiding secret messages by embedding them into digital media while steganalysis is the art and science of detecting the hidden messages. The goal of a high quality steganography is hiding information imperceptibly not only to human eyes but also to computer analysis. The obvious purpose of steganalysis is to collect sufficient evidence about the presence of embedded message and to break the security of the carrier. Steganalysis can be seen as a pattern recognition problem also since based on whether an image contains hidden data or not, images can be classified into Stego or Cover image classes. Steganalysis is broadly classified into two categories. One is meant for breaking a specific steganography. The other one is universal steganalysis, which can detect the existence of hidden message without knowing the details of steganography algorithms used. Universal steganalysis is also known as blind steganalysis and it is more applicable and practicable [1,2] than the specific steganalysis. Based on the methods used, steganalysis techniques are broadly classified into two classes; signature based steganalysis and statistical based steganalysis. Specific signature based steganalysis are simple, give promising results when message is embedded sequentially, but hard to automatize and their reliability is highly questionable [3,4]. The first blind steganalysis algorithm to detect embedded messages in images through a proper selection of image quality metrics and multivariate regression analysis was proposed by Avcibas