CASIA OpenIR  > 毕业生  > 硕士学位论文
Alternative TitleSubspace Analysis Method for Face Recognition
Thesis Advisor马颂德
Degree Grantor中国科学院研究生院
Place of Conferral中国科学院自动化研究所
Degree Discipline模式识别与智能系统
Keyword人脸识别 子空间分析 主元分析 线性判别分析 零空间 基于核技术的子空间分析 Face Recognition Subspace Analysis Principal Component Analysis Linear Discriminant Analysis Null Space Kernel-based Subspace An
Abstract人脸识别作为计算机视觉、模式识别和图像处理等领域研究的重要应用之 一,近年来越来越受到研究人员的关注。与其他生物特征识别方法比较,人脸识 别的优点在于自然、友好、无侵犯性,并且相对成本较低,但是识别的准确率还 有待进一步提高。 子空间分析方法是统计模式识别中一类重要的方法,其基本思想是寻找高维 原空间(样本空间)的一个低维子空间(特征空间),将原样本投影到这个子空 间上再进行分类。这样一方面对高维样本进行了降维、压缩,大大简化了计算; 更重要的是,通过正确选择子空间,高维样本在子空间上的投影可以比在原空间 中具有更好的可分性,使分类更准确。不同的子空间分析方法遵循不同的准则, 可以得到不同的子空间。在本文中我们将详细介绍各种子空间分析方法(例如主 元分析方法,线性判别分析方法,独立元分析方法,以及基于核技术的子空间分 析方法等)在人脸识别中的应用。 我们的主要工作在于对不同的子空间方法进行了比较和评价;在已有的利用 零空间的线性判别分析方法的基础上提出了一种简化的零空间方法来解决线性 判别分析中的小样本问题;以及将基于核技术的线性判别分析方法应用于人脸识 别并取得了较好的效果。 我们将在第一章中介绍人脸识别以及相关的生物特征识别的基本概念,人脸 识别问题的研究意义,研究困难,评价标准,并介绍广义人脸识别系统的组成。 第二章简单综述了当前有代表性的人脸识别方法。 第三章详细介绍各种子空间分析方法及其在人脸识别中的应用,重点是我们 的工作。 第四章是对全文的总结及对人脸识别研究前景的展望。
Other AbstractAs one of the most important applications of Computer Vision, Pattern Recognition and Image Processing, Face Recognition has recently received more and more extensive attention. Compared with other Biometrics, Face Recognition technology is more acceptable because it is more natural, friendly and non-intrusive, and its cost is relatively low, but the accuracy of Face Recognition is not high enough at present. Subspace Analysis is one of important methods of Statistical Pattern Recognition. The basic idea of Subspace Analysis is to find a subspace (feature space) of the original space (sample space), and classify the projected samples in the subspace. One advantage of this method is that the computational complexity is reduced because of the dimensionality reduction of the samples. More importantly, the classification of the samples will be easier in the subspace if we choose an appropriate subspace, so the accuracy ~f recognition is higher. We can get different subspace by using different criteria. In this paper we will introduce various Subspace Analysis methods (such as Principal Component Analysis, Linear Discriminant Analysis, Independent Component Analysis, and Kernel-based Subspace Analysis etc.) and their applications in Face Recognition. The main contribution of this paper is that various Subspace Analysis methods are compared and evaluated; a new simplified method based on the null space is proposed to solve the Small Sample Size Problem of Linear Discriminant Analysis; and Kernel-based Linear Discriminant Analysis method is used in Face Recognition and gets good performance. Some concepts about Face Recognition and Biometrics will be in- troduced in Chapter 1, followed by the significance, difficulties and evaluation criteria of Face Recognition research. Also an ideal Face Recognition system architecture is outlined in Chapter 1. Chapter 2 is a brief survey on representative Face Recognition methods. Chapter 3 introduces in detail various Subspace Analysis methods, especially our work, and their applications in Face Recognition. Chapter 4 summary the whole paper and give a prospect of the de- velopment of Face Recognition technology.
Other Identifier646
Document Type学位论文
Recommended Citation
GB/T 7714
黄锐. 人脸识别中的子空间分析方法[D]. 中国科学院自动化研究所. 中国科学院研究生院,2002.
Files in This Item:
File Name/Size DocType Version Access License
硕士生学位论文-646.pdf(7249KB) 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[黄锐]'s Articles
Baidu academic
Similar articles in Baidu academic
[黄锐]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[黄锐]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.