基于加权小波变换和FLD的表情特征提取  

Expression feature extraction based on weighted wavelet transform and FLD

作　　者：喻丽春 刘金清 YU Li-chun;LIU Jin-qing(Department of Information Technology,Fuzhou University of International Studies Fujian and Trade,Fujian Fuzhou 350202,China;College of Photonic and Electronic Engineering,Fujian Normal University,Fujiang Fuzhou 350007,China)

机构地区：[1]福州外语外贸学院信息系,福建福州350202 [2]福建师范大学光电与信息工程学院,福建福州350007

出　　处：《贵阳学院学报（自然科学版）》2017年第4期71-74,共4页Journal of Guiyang University：Natural Sciences

基　　金：国家自然科学基金资助项目:"仿真视觉系统多通道并行异构神经网络的目标识别算法研究"(项目编号:61179011);福建省自然科学基金资助项目:"医学图像分割算法的研究"(项目编号:2010J01327)

摘　　要：为了解决表情特征变化引起的人脸识别率降低问题,设计了一种人脸表情特征提取方法。首先对人脸图像进行小波变换,将四个小波子带进行加权融合,接着采用主成分分析法(principal component analysis,PCA)将融合后的图像映射到低维空间,最后使用线性判别分析法(Fisher Linear Discriminant,FLD)方法进行表情特征提取,仿真实验表明,该方法能有效克服表情姿势变化对人脸识别的影响,提高了人脸识别率和识别速度。In order to solve the problem of facial recognition rate reduction caused by facial feature change,a facial expression extraction method is designed.Firstly,the wavelet transform of the face image,with four wavelet weighted fusion,then using PCA to map images fused to low dimensional space,finally using Linear Discriminant analysis method of expression feature extraction.Simulation results show that this method can effectively overcome the influence of expression and gesture recognition and improve the recognition rate and the speed of recognition.

关 键 词：小波变换 特征脸 主成分分析 FISHER线性判别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]