Effective steganalysis based on statistical moments of wavelet characteristic function

Abstract

In this paper, an effective steganalysis based on statistical moments of wavelet characteristic function is proposed. It decomposes the test image using twolevel Haar wavelet transform into nine subbands (here the image itself is considered as the LL 0 subband). For each subband, the characteristic function is calculated. The first and second statistical moments of the characteristic functions from all the subbands are selected to form an 18-dimensional feature vector for steganalysis. The Bayes classifier is utilized in classification. All of the 1096 images from the CorelDraw image database are used in our extensive experimental work. With randomly selected 100 images for training and the remaining 996 images for testing, the proposed steganalysis system can steadily achieve a correct classification rate of 79% for the non-blind Spread Spectrum watermarking algorithm proposed by Cox et al., 88% for the blind Spread Spectrum watermarking algorithm proposed by Piva et al., and 91% for a generic LSB embedding method, thus indicating significant advancement in steganalysis.

Steganalysis of LSB matching based on statistical modeling of pixel difference distributions

“…Ker [9] Calibrated adjacency HCF-COM 1 Lie et al [10] Gradient energy and variance of the Laplacian parameter 2 Zhang et al [21] The sum of the absolute differences between local extreme and their neighbors in the intensity histogram 1 Huang et al [8] The alteration rate of the number of neighborhood gray-levels 1 Holotyak et al [7] Higher-order statistical features derived from an estimation of the stego-signal in the wavelet domain 33 Goljan et al [5] Higher-order absolute moments of the noise residual 27 Shi et al [18] The moments of the characteristic functions (CF) of wavelet coefficients and prediction-error images 78 Shi et al [19] The moments of the CF of wavelet coefficients 18 Liu et al [11] The COM of HCF of high-order differential histograms and joint distribution histograms 30 Liu et al [12] Complexity measure, entropy, and high-order statistics of the histogram of the nearest neighbors, probabilities of the equal neighbors, and correlation features 118 Liu et al [13] Bit-plane auto-correlation measurements 54 Marvel et al [14] Statistics calculated from Bit-plane Context Tree Weighting (BTCW) image compressor predications 12 Pevný et al [16] Subsets of sample transition probability matrices 686…”

Passive steganalysis based on higher order image statistics of curvelet transform

“…Holotyak et al [31] used higher-order moments of the PDF of the estimated stego-object in the finest wavelet level to construct the feature vectors. Due to the limited number of features used in the steganalysis technique proposed in [30], Shi et al [32] proposed the use of statistical moments of the characteristic functions of the wavelet sub-bands. Because the n-th statistical moment of a wavelet characteristic function is related to the n-th derivative of the corresponding wavelet histogram; the constructed 39-dimensional (39D) feature vector has proved to be sensitive to embedded data.…”

Mixed Norm Regularized Discrimination for Image Steganalysis

“…As the counterpart of steganaography, steganalysis [1] focuses on detecting, extracting or deterring covert communications. Currently, the purpose of image steganalysis is to detect the presence of hidden messages in cover photographic images.…”