东篱科研大数据发现系统（DRDS）

位置：成果数据库 > 期刊 > 期刊详情页

基于对称ICA的特征抽取方法及其在人脸识别中的应用

ISSN号：1003-6059
期刊名称：《模式识别与人工智能》
时间：0
分类：TP391.41[自动化与计算机技术—计算机应用技术;自动化与计算机技术—计算机科学与技术]
作者机构：[1]南京理工大学计算机科学与技术系,南京210094, [2]江苏科技大学电子信息学院,镇江212003, [3]中国科学院沈阳自动化研究所机器人学重点实验室,沈阳110015
相关基金：国家自然科学基金（No.60472060,60572034）、南京理工大学科研发展基金（No.AB96125）资助项目

关键词：独立成分分析, 对称独立成分分析, 人脸识别, 奇偶分解, 镜像变换, Independent Component Analysis, Symmetrical Independent Component Analysis, Face Recognition, Even/Odd Decomposition, Mirror Transform

中文摘要：

独立成分分析在信号处理和图像处理领域已受到广泛关注．本文提出了一种基于人脸直观自然特性的新算法——对称独立成分分析．该算法在代数上基于函数分解的思想．几何上基于人脸的镜像对称．首先引人镜像变换，然后根据奇、偶分解原理，分别生成奇偶对称样本，最后分别提取各奇偶样本的独立成分．理论分析与实验证明，该算法巧妙利用了镜像样本，既扩大了样本容量，又提高了识别率．同时该算法对光照变化有一定的不敏感性．

英文摘要：

Independent Component Analysis （ ICA ） has been extensively used in the field of signal processing and image processing. In this paper, a new algorithm called symmetrical ICA （SICA） based on facial symmetry is proposed. This algorithm is based on the theory of [unction decomposition in algebra and mirror symmetry in geometry . In this algorithm , mirror transform is firstly introduced. Then, even/odd symmetrical samples are produced based on the theory of the even/ odd decomposition principle, and the even/odd independent components are extracted from the corresponding samples respectively Both theoretical analysis and experimental results demonstrate that this algorithm not only enlarges the number of training samples , but also remarkably raises the recognition rate. Experiment results also show this algorithm is not sensitive to the illumination variation of human faces.