摘要人脸数据处理经常利用数据分析工具,以发现数据中隐藏的各种信息。典型相关分析(CCA)作为一种研究两组变量之间相关关系的多元统计方法,能较好地揭示出两组变量之间的内在关系。近年来,CCA已被广泛应用于多组特征间的融合与抽取,并在生物特征识别与图像分析领域取得了系列研究成果。CCA的基本思想是首先建立两组变量之间的相关性判别目标函数,通过最优化该目标函数求解两组相关投影矢量集,以抽取各个变量的相关性特征,然后在不同的特征融合策略下获得组合的相关特征,最后将其用于图像的分类。47853
本文以CCA的数学模型为研究对象展开拓展研究,提出了基于CCA的改进算法,并通过人脸数据库上的实验验证了其有效性及稳定性。首先,本文基于对人脸识别理论的学习,完善了CCA用于组合特征抽取及人脸识别的理论框架,提出一种新的特征融合策略;其次,本文在研究核CCA及广义CCA的基础上,建立了核广义CCA的准则函数,并通过数学推导对其进行了模型求解和算法设计;最后,本文基于相关系数的几何意义,提出了一种新的基于旋转变换的CCA算法模型,并讨论了模型的几何意义。
关键词 人脸识别 特征抽取 典型相关分析 核方法 旋转变换
毕业设计说明书(论文)外文摘要
Title Research on Feature Fusion Algorithm Based on Improved Canonical Correlation Analysis
Abstract
In the processing of face images, we usually use the data analysis tools to find the underlying relations in data. As an extraction method based on two groups of features,Canonical Correlation Analysis(CCA) can better reveal the inner relations between two variates. Recently, CCA has been widely used in multiple feature fusion and extraction and it has achieved a series of achievements.The basic idea of CCA is: firstly, establish correlation criterion function between two groups of features; then extract the correlation feature of each data; finally, obtain the combined correlation feature, which is used in image classification.
In this paper, we focus on the study of CCA and propose several improved algorithms . Firstly, we discuss the framework of CCA used in combined feature extraction and face recognition. Besides,we propose a new feature fusion strategy. Secondly, we apply the kernel idea to generalized CCA(GCCA)and derive the solution of this model. Finally, on the basis of geometric meaning of correlation coefficient, we propose a new algorithm of CCA on conversion of coordinates and we discuss the geometric meaning of this model.
Keywords face recognition , feature extraction , Canonical Correlation Analysis , kernel method , rotation transformation
目 录 I
1 绪论 1
1.1 课题研究背景 1
1.2 典型相关分析的研究与发展 2
1.3 本文主要内容及工作安排 3
1.4 本文的主要创新点 3
2 典型相关分析模型及特征融合 5
2.1 典型相关分析模型 5
2.2 核典型相关分析 8
2.3 一种新的特征融合策略 10
2.4 实验与分析 11
2.5 本章小结 15
3 核广义典型相关分析的算法研究 基于改进的典型相关分析的特征融合算法研究:http://www.youerw.com/jisuanji/lunwen_50115.html