【摘要】 在数字水印应用中,利用模拟函数将秘密信息编码为载体特征值,并使特征值在编码与原始载体中的概率分布一致,能提高掩密载体统计上不可检测的性能.采用相对熵作为度量随机变量两种分布距离的指标,设计并证明了具有任意精度的模拟函数计算方法.根据指定精度将特征值变量的概率分布用幂级数扩展,构造前缀码,利用前缀码对秘密信息编码得到模拟结果.数值实验也证实了该算法的有效性.更多还原

【Abstract】 In the applications of digital watermark,mimic funtion encodes secret messages to features of cover subject according to a given probability distribution the features assumes before,and thus keeps the stego-subject imperceptible statistically.Taken relative entropy as a measure of difference between two probability distributions of one random variable,a mimic function which has any fixed precision was designed and proved.To decrease the relative entropy as possible,the probability space of features of cover subject was firstly expanded by decomposing each feature to several temporary symbols,each symbol was allotted a probability of power of a given integer,and the summation of these powers approximate the probability of corresponding feature in original space,repectively.Then a prefix code was generated based on the new space and used to convert secrete messages to mimic function value.The effection of the function designed here was also validated by experimental result.更多还原