基于保守超混沌系统的虹膜特征提取算法  

Iris Feature Extraction Algorithm Based on Conservative Hyperchaotic System

作　　者：赵中军[1] 杜久龙 邓雷升[1] 涂清 ZHAO Zhongjun;DU Jiulong;DENG Leisheng;TU Qing(No.30 Institute of CETC,Chengdu Sichuan 610041,China)

机构地区：[1]中国电子科技集团公司第三十研究所,四川成都610041

出　　处：《通信技术》2024年第2期132-138,共7页Communications Technology

摘　　要：虹膜特征与其他生物特征相比具有独特优势,因其可以为加密算法生成足够长的密钥。虹膜特征提取是虹膜识别过程中的关键问题。针对密钥安全性和计算效率两大问题,提出了一种新颖的基于保守超混沌系统的虹膜特征提取方法,并结合局部二进制模式(Local Binary Pattern,LBP)和高斯函数差分(Difference of Gaussian,DOG)技术。首先,选择五维保守超混沌系统,验证其保守性和超混沌性质;其次,根据375位虹膜特征码和上述系统,产生特征矩阵,并对该矩阵进行图像增强处理;最后,通过提取DOG极值点和LBP特征,并根据提出的特征融合规则生成128位密钥。实验表明,生成的密钥可以通过美国国家标准与技术研究院(National Institute of Standards and Technology,NIST)随机性测试,且对虹膜特征码极其敏感。The iris feature has unique advantages over other biometric features because it can generate sufficiently long keys for encryption algorithm.Iris feature extraction is a key issue in iris recognition process.In response to the two major issues of key security and computational efficiency,a novel iris feature extraction method based on conservative hyper-chaotic system is proposed,which combines LBP(Local Binary Pattern)and DOG(Difference of Gaussian)technology.First,a five-dimensional conservative hyper-chaotic system is selected and its conservatism and hyper-chaotic properties are verified.Then,according to 375-bit iris feature code and the above system,a feature matrix is generated and the matrix is subjected to image enhancement processing.Finally,DOG extreme points and LBP features are extracted from the above matrix,and 128 final keys are generated according to the feature fusion rules proposed in this paper.The experimental results indicate that the generated keys can pass the NIST(National Institute of Standards and Technology)randomness test and are extremely sensitive to iris feature codes.

关 键 词：超混沌系统 保守系统 虹膜特征提取 局部二进制模式LBP 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]