基于互信息的高斯指纹方案  

Gaussian fingerprinting scheme based on mutual information

作　　者：程格平[1] 

机构地区：[1]湖北文理学院数学与计算机科学学院,湖北襄阳441053

出　　处：《计算机时代》2014年第8期40-42,共3页Computer Era

基　　金：国家自然科学基金项目(60873226)

摘　　要：在高斯指纹系统中,指纹的嵌入与检测须服从均方失真限制,以获得较好的保真度。根据概率统计理论,分析了基于阈值相关检测的传统指纹方案中指纹编码速率、均方失真限制以及共谋人数之间的关系,并指出阈值相关检测方法的缺点,即当编码速率大于容量范围的某个值时检测器性能较差。为了解决容量限制的问题,提出一种在限定均方误差的条件下取得数字指纹基本容量的高斯指纹方案。利用互信息游戏理论,提出最大惩罚高斯互信息的指纹检测方法,有效地解决了传统指纹检测方法存在的问题。根据指纹容量的数学模型,推导出指纹容量的表达式。In Gaussian fingerprinting system, fingerprinting encoder and decoder must obey mean-squared distortion constraints to obtain better fidelity. According to the theory of probability and statistics, the fingerprinting code rate is analyzed with mean-squared distortion on constraint and the number of colluders on account of the normalized threshold decoder. The disadvantage of traditional fingerprinting decoding method by correlation decoder which has poor decoding performance when code rates greater than some value of capacity is introduced. To solve this problem, a Gaussian fingerprint scheme is proposed to achieve the fundamental capacity limits of digital fingerprint under the definite mean-squared distortion. By means of mutual-information game, the decoding method of maximum penalized Gaussian mutual information is presented. In the end, the expression of fingerprinting capacity is derived from the mathematical model.

关 键 词：高斯指纹 互信息理论 指纹检测 指纹容量 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]