Uncorrelated Multiple Information Projection Based on Locality Preserving Projections
|更新时间:2026-07-10
|
Uncorrelated Multiple Information Projection Based on Locality Preserving Projections
Vol. 15, Issue 5, Pages: 75-79(2010)
作者机构:
1. 哈尔滨理工大学计算机科学与技术学院
2. 伊春职业学院机电工程系
作者简介:
基金信息:
DOI:
CLC:TP391.41
Published:2010
稿件说明:
移动端阅览
LIN Ke-zheng, CHEN Dong-mei, WU Qing-ying. Uncorrelated Multiple Information Projection Based on Locality Preserving Projections[J]. 2010, 15(5): 75-79.
DOI:
LIN Ke-zheng, CHEN Dong-mei, WU Qing-ying. Uncorrelated Multiple Information Projection Based on Locality Preserving Projections[J]. 2010, 15(5): 75-79.DOI:
Uncorrelated Multiple Information Projection Based on Locality Preserving Projections
摘要
为解决在人脸识别领域的特征提取问题
提出了一种基于局部保持投影(Locality Pre-serving Projections
LPP)的统计不相关复合信息投影(UMIP).该方法是基于线性投影的子空间方法
将原始人脸图像看作是一个矩阵
通过相应的代数方法
抽取得到保持原始样本分布信息的低维子空间信息.在UMIST和Yale标准人脸图像库上的实验结果表明
UMIP算法提高了识别率.
Abstract
With the purpose of solving feature extraction problem in face recognition area
a new manifold learning algorithm is proposed
called Uncorrelated Multiple Information Projection(UMIP) based on Locality Preserving Projections.It is an subspace method based on linear projective maps
regards the original picture of face as a matrix
and extracts the lower dimension subspace’s information which optimally preserves the local structure and the global information of the data set by corresponding algebraic method.Th...