Novel Approach of Universal Image Steganalysis using Features Derived from Gray Scale and Color Planes

Madhavi B. Desai; Prof. N. M. Patel; Vipul H. Mistry

Art of hidden communication has evolved a lot with the advancement in technology. It will alter some of the media properties that may introduce few degradation or unusual characteristics. These characteristics may act as signatures that broadcast the existence of the embedded message, thus defeating the purpose of steganography.There are basically two approaches of steganalysis, one is specific approach and other is universal approach. As there are so many data hiding methods, specific steganalysis method is meaningless. That’s why we need to develop a method that can defeat any steganography method and detect the existence of the hidden message. The first and most important stage of steganalysis is feature extraction. In this paper, we have used gray scale and color properties of an image to detect the existence of hidden message. Statistical moments of wavelet characteristic function, prediction error image and energy derived from R, G and B plane of image reflect the change in image properties. We have extracted the features derived from these features and proposed 39-D feature vector to train the SVM classifier. We have tested our proposed method against popular JPEG steganographic methods, spread spectrum and LSB like method. Our proposed method is able to achieve average 85% detection rate against all steganographic methods. We have used standard image database of 100 CorelDraw images of size 256x256 for our experimental purpose. This paper also compares the performance of proposed method with other existing steganalysis methods.

Select Volume / Issues:

• Volume 3 Issue 6, Nov. 2012

Year:

2012

Type of Publication:

Article

Keywords:

Universal Steganalysis; Cover Image; Stego Image; CF Moments; Prediction Error Image; Bayes Classifier; SVM Classifier

Journal:

IJECCE

Volume:

3

Number:

6

Pages:

1620-1625

Month:

November