摘要
局部线性嵌入是一种当今比较流行的流形学习算法,该算法能够实现对数据的有效降维。人脸表情有着复杂的纹理和形状特征,提取出的数据往往存在着大量的数据冗余,因此数据降维在人脸表情识别中有着重要的作用。本文针对局部线性嵌入对分布不均匀的数据不能有效降维的问题提出了改进算法,并应用在人类表情分类中,通过仿真实验分析,改进算法对人类表情图像具有良好的降维和分类效果。
Locally Linear Embedding is one of the popular manifold learning algorithm,which is good at data dimensionality reduction.The facial expression has complex shape and texture features,and there are a huge of data redundancy,so data dimension reduction plays an important role in facial express recognition. But,Local Linear Embedding can not achieve good dimension reduction result for uneven data. Based on this,we propose a LLE improved algorithm and practise it in facial expression classification. By analysis of simulation experiment,we can conclude that the improved algorithm can receive better results.
出处
《北华航天工业学院学报》
CAS
2014年第3期16-19,共4页
Journal of North China Institute of Aerospace Engineering
基金
廊坊市科技计划项目(2013011008)
关键词
局部线性嵌入
表情识别
降维
local linear embedding
facial expression recognition
dimension reduction
