On the Possible Detectability of Image-in-Image Steganography

Abstract

This paper investigates the detectability of popular imagein-image steganography schemes [1, 2, 3, 4, 5]. In this paradigm, the payload is usually an image of the same size as the Cover image, leading to very high embedding rates. We first show that the embedding yields a mixing process that is easily identifiable by independent component analysis. We then propose a simple, interpretable steganalysis method based on the first four moments of the independent components estimated from the wavelet decomposition of the images, which are used to distinguish between the distributions of Cover and Stego components. Experimental results demonstrate the efficiency of the proposed method, with eight-dimensional input vectors attaining up to 84.6% accuracy. This vulnerability analysis is supported by two other facts: the use of keyless extraction networks and the high detectability w.r.t. classical steganalysis methods, such as the SRM combined with support vector machines, which attains over 99% accuracy.

Figures (6)

• Figure 1: Overview of the proposed steganalysis method. First, a discrete wavelet transform (DWT) is applied to the input image. Then, principal component analysis (PCA) is performed to extract weak components. Independent component analysis (ICA) is applied to extract independent sources that are semantically close to the Cover and Payload images. Finally, the first four moments (mean $\mu$, standard deviation $\sigma$, skewness $\gamma$, and kurtosis $\kappa$) of the estimated independent components are computed to form a discriminative yet interpretable feature vector, which is used to train an SVM classifier for detection.
• Figure 2: Illustration of the general architecture of the coupling layer of the INN in the HiNet model jing2021hinet. The complete INN is formed of 16 blocks. The arrows indicate the direction of the forward pass.
• Figure 3: Secret image from a HiNet Stego image using different noise inputs. From left to right: Stego image, Payload image, revealed Payload with random noise, revealed Payload with zero noise. The PSNR (in dB) is computed between the revealed secret images and the original secret image.
• Figure 4: Embedding changes in each sub-band of the DWT of the stego image, computed as the difference with the DWT sub-bands of the cover.
• Figure 5: Correlation matrix between the sub-bands of the DWT of the payload image and the difference between the ones of the Cover and the Stego images, when using the HiNet model jing2021hinet. The Pearson correlation coefficient is used. It shows that the low-frequency sub-bands of the payload are mostly embedded in the Stego image.