【作者】 董火明；

【导师】 高隽；

【作者基本信息】 合肥工业大学 ， 信号与信息处理， 2004， 硕士

【摘要】 模式识别的协同学方法自Haken教授提出以来，作为一种崭新的模式识别方法，人们进行了广泛的理论与应用研究，协同识别理论已被成功的应用于模式识别等诸多领域。生物识别技术被认为是终极的身份识别手段，但目前的研究还处于初始发展阶段，如何提高系统的识别速度、准确性和鲁棒性是这个研究领域的重大挑战。 本文对协同识别理论进行了较为深入的研究，进而将协同识别理论及其研究成果应用于生物识别之中，针对现有生物识别技术存在的不足，开展指纹与人脸的全局特征识别与身份认证的应用研究。 本文的主要工作及成果如下： ● 介绍了协同学的基本思想、数学模型和关键概念，并在协同学的基本方程的基础上，详细的阐述了协同识别理论，包括其数学模型、网络实现、关键概念和识别过程及关键技术等内容； ● 在总结和整理的基础上，对协同识别理论进行了较为系统的探讨；提出了一种基于K-均值的聚类型协同学习算法；提出了一种基于PCA特征提取的协同识别方法，是基于全局特征的模式识别方法；对协同识别中的参数进行了深入的理论分析，并提出了一种基于模拟退火机制的参数优化方法； ● 综述了生物识别技术，以指纹识别和人脸识别为重点，较为系统的介绍了其发展历史、主要方法与关键技术、以及其应用领域与发展趋势等内容； ● 指出了基于全局特征的生物识别技术的研究意义，以及协同识别理论应用于生物识别中的思路与结合点；提出了基于全局特征的协同指纹识别和协同人脸识别方法，并利用多分类器融合及其改进的方法，来改善生物识别的效果。更多还原

【Abstract】 Since Prof. Haken introduced synergetics methods to pattern recognition,lots of researches have been kept on investigating. Synergetic recognition theory has been applied to pattern recognition and many other fields. Biometrics techniques are considered as the best person authentication method,but there is lots of work to be done. How to improve the recognition speed,veracity and robustness of the system is still the challenge in the fields.This dissertation deals with the researches on synergetic recognition theory,and applies the theory research work to biometrics. Aiming at the shortages of biometrics methods in existence,we present novel approaches of fingerprint and face recognition for person authentication based on global features.The main work and contributions of the dissertation are summarized as follows:Firstly,we introduce the fundamental principle,mathematics model and key concepts of synergeics. On the basis of the principal equation of synergeics,we detail synergetic recognition theory,including the mathematics model,network realization,key concepts,recognition process and key techniques.Secondly,we systematically discuss the synergetic recognition theory. And then we propose a novel synergetic learning algorithm based on K-mean cluster,and present a new synergetic recognition approach based PCA features extraction,which is a pattern global features recognition method. We also abstractly analyze the parameters of synergetic recognition approach,and purpose a novel parameters optimization method based on simulated annealing mechanism.Thirdly,we introduce biometrics techniques,especially fingerprint recognition and face recognition in detail. We discuss the following content such as their development history,main methods and key techniques,and application fields and development current.Finally,we point out the importance of biometrics based global features,and the combination idea with synergetic recognition theory. Then,we purpose synergetic fingerprint recognition and synergetic face recognition method based on global features,and improve the biometrics recognition performance using multiple classifiers fusion approach.更多还原