基于随机分布的图像域隐写分析算法研究

Research on Steganalysis Algorithm of Image Domain Based on Random Distribution

信息隐藏是信息安全研究的重点领域.随着隐写术的广泛应用,网络中的攻击者通过隐写来进行隐蔽通信传递信息,造成了极大的安全隐患,因此作为隐写对抗的隐写分析技术变得尤为重要.本文提出了一种基于Fisher-Yates隐写数据随机置乱算法,首先利用BOSSBase 1.01数据集,对nsF5、J-UNIWORD和UERD 3种隐写算法进行训练,再通过隐写分析残差网络(SRNet)对9种不同嵌入率的数据集进行隐写分析,通过对隐写算法和隐写分析做对抗测试,验证了随机分布的隐写分析残差网络在图像域的有效性.

Information hiding is a crucial area of information security research.With the wide application of steganography, the attackers in the network use steganography to communicate and transmit information, causing great security risks.Therefore, steganography analysis technology as a steganography confrontation becomes particularly important.This paper proposes a random scrambling algorithm based on Fisher-Yates steganographic data.First, the BOSSBase 1.01 data set is used to train three steganographic algorithms, nsF5,J-UNIWORD,and UERD,and then the steganalysis residual network(SRNet) is used to carry out steganalysis on nine datasets with different embedding rates.Through confrontation tests on steganalysis algorithms and steganalysis, this paper verifies the effectiveness of the stochastically distributed steganalysis residual network in the image domain.