文档介绍：上海交通大学硕士学位论文
2) 介绍分析各种常用降维算法的原理和关系。这里本论文主
要关注基于谱分析的算法,即用矩阵特征分解来求解的算法。我们介
绍的算法包括传统的算法:PCA 和 LDA;流形学习算法:LLE、LPP、
ISOMAP、LE、LTSA 和 HLLE。
3) 基于子空间降维方法的步态识别研究。提出了一种通过统
计行走模式动态信息的步态识别方法。对代表每一类的周期平均侧影
图像(GEI)进行方差分析求得动态权值掩模(DWM)。通过 DWM 对原始
GEI 进行动态和形状信息的增强以获得新的步态表征 EGEI。为增加可
辨识信息,使用一组 Gabor 小波对 EGEI 进行卷积,然后采用辨别共
同向量分析(DCV)将高维卷积结果在低维空间表示;提出了一种有
效的基于步态能量图像的身份识别方法。首先生成合成 GEI 丰富训练
集样本数量。然后利用在以前文献中被忽略的具有良好识别性能的
Gabor 相位信息作为身份特征,并采用流型学习算法保局影射(LPP)
将此高维数据在低维空间表示。通过使用简单的分类策略在 USF 步态
数据库上的对比实验,证明了提出的两种方法对识别性能提高的有效
性。
4) 基于视频的实时人脸身份识别方法。将人脸检测和跟踪、
姿态估计和人脸识别纳入视频监控系统。系统首先检测到候选人脸,
并实现稳健的跟踪,然后对人脸区域进行预判断,识别出人脸的朝向,
如果是正面朝向就是对该人脸进行身份判断。相对于传统的静态识别
方案,系统可以实现无需初始化、无需主动配合、全自动实时识别。
具有更高的效率和灵活性。
关键词:生物特征识别,降维,视频监控,步态识别,人脸姿态估计,
人脸识别
II
上海交通大学硕士学位论文
Biometric Applications based on
Subspace Dimensional Reduction Algorithms
Abstract
A wide variety of systems requires reliable personal recognition
schemes to either confirm or determine the identity of an individual
requesting their services. The purpose of such schemes is to ensure that
the rendered services are accessed only by a legitimate user and no one
else. Examples of such applications include secure access to buildings,
computer systems, laptops, cellular phones, and ATMs. Traditional
identity recognition systems operate by “what she possesses”(., an ID
card) or “what she remembers”(., a password). These modes are
vulnerable to the wiles of an impostor. Biometrics refers to the automatic
recognition of individuals based on their physiological and/or behavioral
characteristics. It confirms or establishes an individual’s identity based on
“who she is”, which is more reliable than traditional ones. A biometric
system may be viewed as a signal detection system with a pattern
recognition architecture that senses a raw biometric signal, processes this
signal to extract a salient set of features, compares these features against
the feature sets residing in the database,