Abstract:Multi-image steganography refers to a data-hiding scheme where a user tries to hide confidential messages within multiple images. Different from the traditional steganography which only requires the security of an individual image, multi-image steganography considers an overall security for a batch of images. However, existing multi-image steganography all faces a nontrivial problem: how to optimally allocate payload into multiple images to guarantee the security of batch images. To address this problem, this paper proposes an adversarial multi-image steganographic scheme. A multi-scale texture evaluation mechanism is firstly calculated to determine the embeddable cover images. Subsequently, a series of multi-scale filters are introduced to enhance the image content, which can be used to guide the optimal payload assignment of each image. Furthermore, an adversarial embedding mechanism is designed by dynamically adjusting the random gradient mapping of batch images, and finally achieving secure multi-image steganography. Our proposed scheme can optimize the overall steganographic security performance of multiple images, while ensuring the anti-steganalysis capability of a single image. Extensive experiments demonstrate that our scheme can achieve superior performance for multi-image steganography over different large-scale image sets, and outperforms state-of-the art schemes in terms of both single-image and multi-image anti-steganalysis capability.

Content-Adaptive Steganalysis Via Augmented Utilization Of Selection-Channel Information

Detection of audio-to-image audio steganography based on peak frequency feature

A Dct-Based Image Steganographic Method Resisting Statistical Attacks

A New Steganographic Method Resisting Statistical Attack

A Blind Steganalytic Scheme Based on DCT and Spatial Domain for JPEG Images.

Steganalysis of model based steganography and steghide in grayscale JPEG images

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

A Content-Adaptive Approach for Reducing Embedding Impact in Steganography

Embedding Probability Guided Network for Image Steganalysis.

A New SDCS-based Content-adaptive Steganography Using Iterative Noise-Level Estimation

Adaptive Steganalysis Based on Statistical Model of Quantized DCT Coefficients for JPEG Images

Improving The Embedding Strategy For Batch Adaptive Steganography

Adaptive Audio Steganography Based on Improved Syndrome-Trellis Codes

Enhancing reliability and efficiency for real-time robust adaptive steganography using cyclic redundancy check codes

A New Cost Function for Spatial Image Steganography Based on 2D-SSA and WMF

Double-Layered Dual-Syndrome Trellis Codes Utilizing Channel Knowledge for Robust Steganography

Robust Adaptive Steganography Based on Dither Modulation and Modification With Re-Compression

Maximizing complex features to minimize the detectability of content-adaptive steganography

An Effective Imbalanced JPEG Steganalysis Scheme Based on Adaptive Cost-Sensitive Feature Learning

A Markov Process Based Approach to Effective Attacking JPEG Steganography

Adversarial multi-image steganography via texture evaluation and multi-scale image enhancement