Abstract: This paper analyzes the theoretical security model which is proposed by Christian Cachin, and applies this model to evaluate the statistical imperceptibility of information hiding. This evaluating method removes correlation by blocking image, and uses Discrete Cosine Transform(DCT) to get the image’s stationary probability distributions. It computes DCT coefficients’ joint probability distributions to evaluate information hiding algorithms’ statistical imperceptibility. Results of experiments indicate that this method not only can be good for evaluating statistical imperceptibility, but also can guide information hiding algorithm’s design.

Key words: information hiding, security, relative entropy, joint probability distributions, statistical imperceptibility

摘要： 对Cachin理论安全性进行分析，应用该模型对信息隐藏的统计不可见性进行评价。通过将图像分块去相关性，运用DCT变换得到载体的平稳概率分布，计算DCT系数的联合概率分布的相对熵来评估隐藏算法的统计不可见性。实验结果表明，该评估方法不仅能有效衡量隐藏统计不可见性，还能用以指导隐藏算法的设计。

关键词: 信息隐藏, 安全性, 相对熵, 联合概率分布, 统计不可见性

CLC Number: 

• TP309