【作者】 刘娅；

【导师】 李春明；

【作者基本信息】 河北科技大学 ， 测试计量技术及仪器， 2010， 硕士

【摘要】 信息技术的飞速发展对身份验证的可信度和便捷度提出了更新的要求,一些传统的身份认证方法已经不能满足这些新的需求,用生物特征来进行身份认证已成为研究热点。人脸识别是模式识别技术在图像领域中的具体运用,其应用前景非常广阔,作为一种典型的生物特征识别,人脸识别以其自然性、可接受性、采集方便等优点受到了人们的青睐,可以应用到身份证件的鉴别、自动门禁控制系统、刑侦破案、视频监控、家庭安全等领域。本文首先介绍了现有的人脸识别算法,详细介绍了经典的基于图像向量的PCA人脸识别算法,并针对经典PCA算法在识别当中存在的不足,提出了一种可以提高识别率的PCA算法,即在做算法的识别前加入图像预处理的阶段,这样就能够有效的提高识别率了。然后,在二维主成分分析(2DPCA)的人脸识别算法的基础上提出一种改进的2DPCA的算法(M-2DPCA)。一般情况下,M-2DPCA和2DPCA的识别率接近,高于加入了预处理的PCA算法的识别率。但是M-2DPCA和2DPCA方法在重构时有更高的空间复杂度和时间复杂度,因此重构时要占用更多的内存,当实际应用中系统不能提供足够大内存的情况下,我们就优先选择加入图像预处理的PCA算法。综上所述,本文实现了基于改进的PCA的人脸识别系统,在系统实现的过程中采用了加入图像预处理的PCA方法,使系统的识别率有所提高。本文所设计的身份验证系统,除了对算法的优化外,更加重视系统设计,可更好的发挥算法的效能。更多还原

【Abstract】 Today in the rapid development of information technology, we are made the updated requirements in the authenticated degree of credibility and convenience. Some of the traditional authentication methods clearly can not meet these new demands, and it has become a research hotspot by using biometrics for identity authentication.Face recognition is a pattern recognition technology in the concrete application of the image in the field, and it is very popular in its application prospect . It has been collected to facilitate the people’s favoras as a typical biometric identification for its naturalness, acceptability, convenience. Identity documents can be applied to the identification, automatic access control systems, forensic detection, video surveillance, the security of home and other fields.This article describes the existing face recognition algorithm, with describing the classic image-based vector PCA face recognition algorithm in detail. Because the classical PCA algorithm identifies those shortcomings,we are presenting an improved method of M-PCA. Before the image pre-processing stage ,the recognition algorithm is done,so that we can effectively improve the recognition rate. Then, an improved 2DPCA algorithm (M-2DPCA), M-2DPCA and 2DPCA recognition rate is close to but higher than the M-PCA recognition rate. Under normal circumstances , M-2DPCA and 2DPCA recognition rate are close to each other.However, M-2DPCA and 2DPCA methods remodeling have a higher degree of spatial complexity and time complexity, so we have to spend more time in the reconstruction of memory. When the practical application of the system can not provide enough memory cases, the superiority of PCA in performance is manifested.In summary,this paper presents an improved PCA which is based in face recognition system.It uses PCA method in the process of system implementation, mainly because it can effectively improve the system recognition rate.This article is designed authentication system. In addition to the optimization algorithm, it is greater emphasis on system design, in order to play its algorithm performance.更多还原