Stochastic modulation steganography embeds secret message within the cover image by adding stego-noise with a specific probabilistic distribution. No method is known to be applicable to the estimation of stochastic modulation steganography. By analyzing the distributions of the horizontal pixel difference of images before and after stochastic modulation embedding, we present a new steganalytic approach to accurately estimate the length of secret message in stochastic modulation steganography. The pro- posed method first establishes a model describing the statistical relationship among the differences of the cover image, stego-image and stego-noise. In the case of stegoimage-only steganalysis, rough estimate of the distributional parameters of the cover image＇s pixel difference is obtained with the use of the provided stego-image. And grid search and Chi-square goodness of fit test are exploited to estimate the length of the secret message embedded with stochastic modulation steganography. The experimental results demonstrate that our new approach is effective for steganalyzing stochastic modulation steganography and accurately estimating the length of the secret message.
HE Junhui1,2 & HUANG Jiwu1,2 1. School of Information Science and Technology, Sun Yat-sen University, Guangzhou 510275, China
Rendering technology in computer graphics （CG） is now capable of producing highly photorealistic images, giving rise to the problem of how to identify CG images from natural images. Some methods were proposed to solve this problem. In this paper, we give a novel method from a new point of view of image perception. Although the photorealistic CG images are very similar to natural images, they are surrealistic and smoother than natural images, thus leading to the difference in perception. A pert of features are derived from fractal dimension to capture the difference in color perception between CG images and natural images, and several generalized dimensions are used as the rest features to capture difference in coarseness. The effect of these features is verified by experiments. The average accuracy is over 91.2%.
