Abstract:In recent years, researchers have been making great progress in the steganalysis technology based on convolution neural networks (CNN). However, experts ignore the contribution of nonlinear residual and joint domain detection to steganalysis, and how to detect the adaptive steganographic algorithms with low embedding rates is still challenging. In this paper, we propose a CNN steganalysis model that uses a joint domain detection mechanism and a nonlinear detection mechanism. For the nonlinear detection mechanism, based on the spatial rich model (SRM), we introduce the maximum and minimum nonlinear residual feature acquisition method into the model to adapt to the nonlinear distribution of steganography information. For the joint domain detection mechanism, we not only apply the high-pass filters from the SRM for spatial residuals, but also apply the patterns from the discrete cosine transform residual (DCTR) for transformation steganographic impacts, so as to fully capture the interference trace of spatial steganography to transform domain. We also apply a new transfer learning method to improve the model’s performance. That is, we apply the low embedding rate steganography samples to initialize the model, because we think that the method makes the network more sensitive than applying high embedding rate steganography samples to initialize the model. The simulation results also confirm this assumption. Combined with the above improved methods, the detection accuracy of the model for WOW and S-UNIWARD is higher than that of SRM+EC, Ye-Net, Xu-Net, Yedroudj-Net and Zhu-Net, which is about 4 ∼6% higher than that of the optimal Zhu-Net. The results can provide a certain reference for steganalysis and image forensics tasks.

Highly Accurate End-to-end Image Steganalysis Based on Auxiliary Information and Attention Mechanism

A New Steganalysis Approach Based on Both Complexity Estimate and Statistical Filter

Feature-Based Steganalysis for JPEG Images

Image Steganalysis Network Based on Dual-Attention Mechanism

A New Convolutional Neural Network-Based Steganalysis Method for Content-Adaptive Image Steganography in the Spatial Domain

A Novel Technique for Image Steganalysis Based on Separable Convolution and Adversarial Mechanism

Generative adversarial networks-based image steganography with multiscale features integration

Channel Attention Image Steganography With Generative Adversarial Networks

StegaVision: Enhancing Steganography with Attention Mechanism

Enhancing the Security of Image Steganography Via Multiple Adversarial Networks and Channel Attention Modules.

Steganalysis of Image with Adaptively Parametric Activation

Spatial Steganalysis Based on Non-Local Block and Multi-Channel Convolutional Networks.

Spatial Image Steganography Based on Generative Adversarial Network

IAS-CNN: Image Adaptive Steganalysis Via Convolutional Neural Network Combined with Selection Channel

Image Steganalysis Against Adversarial Steganography by Combining Confidence and Pixel Artifacts.

Generative Focused Feedback Residual Networks for Image Steganalysis and Hidden Information Reconstruction

Image Steganalysis Based on Convolutional Neural Network and Feature Selection

Maximizing steganalysis performance using siamese networks for image

Learning Selection Channels for Image Steganalysis in Spatial Domain

Joint Multi-Domain Feature Learning for Image Steganalysis Based on CNN

Enhancing Steganography Detection with AI: Fine-Tuning a Deep Residual Network for Spread Spectrum Image Steganography