State Key Laboratory of Public Big Data, Guizhou University, Guiyang, China.
College of Computer Science and Technology, Guizhou University, Guiyang, China.
PLoS One. 2024 Sep 6;19(9):e0308265. doi: 10.1371/journal.pone.0308265. eCollection 2024.
Steganography, the use of algorithms to embed secret information in a carrier image, is widely used in the field of information transmission, but steganalysis tools built using traditional steganographic algorithms can easily identify them. Steganography without embedding (SWE) can effectively resist detection by steganography analysis tools by mapping noise onto secret information and generating secret images from secret noise. However, most SWE still have problems with the small capacity of steganographic data and the difficulty of extracting the data. Based on the above problems, this paper proposes image steganography without embedding carrier secret information. The objective of this approach is to enhance the capacity of secret information and the accuracy of secret information extraction for the purpose of improving the performance of security network communication. The proposed technique exploits the carrier characteristics to generate the carrier secret tensor, which improves the accuracy of information extraction while ensuring the accuracy of secret information extraction. Furthermore, the Wasserstein distance is employed as a constraint for the discriminator, and weight clipping is introduced to enhance the secret information capacity and extraction accuracy. Experimental results show that the proposed method can improve the data extraction accuracy by 10.03% at the capacity of 2304 bits, which verifies the effectiveness and universality of the method. The research presented here introduces a new intelligent information steganography secure communication model for secure communication in networks, which can improve the information capacity and extraction accuracy of image steganography without embedding.
隐写术，即使用算法将秘密信息嵌入载体图像中的技术，被广泛应用于信息传输领域，但使用传统隐写算法构建的隐写分析工具可以很容易地识别它们。无嵌入隐写术（SWE）可以通过将噪声映射到秘密信息上，并从秘密噪声生成秘密图像，有效地抵抗隐写分析工具的检测。然而，大多数 SWE 仍然存在秘密数据容量小和数据提取困难的问题。基于上述问题，本文提出了一种无嵌入载体秘密信息的图像隐写术。该方法的目的是提高秘密信息的容量和秘密信息提取的准确性，以提高安全网络通信的性能。所提出的技术利用载体特征生成载体秘密张量，在保证秘密信息提取准确性的同时，提高了信息提取的准确性。此外，Wasserstein 距离被用作鉴别器的约束，并且引入了权重裁剪来增强秘密信息的容量和提取准确性。实验结果表明，该方法在 2304 位的容量下可以将数据提取准确性提高 10.03%，验证了该方法的有效性和通用性。本研究提出了一种新的智能信息隐写安全通信模型，用于网络中的安全通信，可以提高无嵌入图像隐写术的信息容量和提取准确性。
