摘要
线性鉴别分析是特征抽取中最为经典和广泛使用的方法之一。基于人脸的一种直观自然特性——镜像对称性,提出一种算法——对称线性鉴别分析。该算法引入镜像变换,生成镜像样本,依据奇偶分解原理,生成镜像奇、偶对称样本,并分别提取各奇偶样本的对称鉴别特征。理论分析与实验证明,该算法合理地利用了镜像样本,既扩大了样本容量,又提高了人脸识别率。
Linear Discriminant Analysis（LDA） is one of the classical and popular methods used for feature extraction. In this paper, a new algorithm called Symmetrical LDA（SLDA） based on frontal facial symmetry is proposed. This algorithm is based on the theory of function decomposition and mirror symmetry. In the algorithm, mirror transform is introduced. Original face samples are decomposed into even symmetrical images and odd symmetrical ones. Even/odd symmetrical discriminant features are extracted from the corresponding samples respectively. Both theoretical analysis and experimental results demonstrate this algorithm not only enlarges the number of training samples, but also remarkably improves the recognition rates.
出处
《计算机工程》
CAS
CSCD
北大核心
2010年第1期201-202,205,共3页
Computer Engineering
基金
国家自然科学基金资助项目(60572034)
江苏省自然科学基金资助项目(BK2004058)
江苏科技大学电子信息学院青年教师科研立项基金资助项目
关键词
人脸识别
镜像对称性
对称线性鉴别分析
face recognition
mirror symmetry
Symmetrical Linear Discriminant Analysis（SLDA）
