摘要
针对非限制性条件下人脸识别准确率较低的问题,提出了一种基于方向梯度直方图特征提取和快速主成分分析算法的人脸识别方法。首先,利用Haar特征分类器对原始数据进行人脸的检测与提取,并进行有序化保存;然后,对有序数据进行方向梯度直方图特征提取,再使用快速主成分分析算法进行降维处理,并对其进行归一化处理;最后,采用支持向量机算法(SVM)对所得到的实验数据进行最终的分类与识别。实验结果表明,与单一的支持向量机算法、主成分分析算法以及方向梯度直方图算法相比,本文方法能有效地提高非限制性条件下人脸识别的准确率,且耗时较短。
To solve the problem of low accuracy of face recognition under non-restrictive conditions,a new face recognition method based on Histogram of Oriented Gradient（HOG）feature extraction and fast Principal Component Analysis（PCA）algorithm is proposed.In this method,firstly the original data is detected and extracted by using the Haar feature classifier,and the extracted data are then stored in order.Secondly,the ordered data are extracted from the HOG feature extraction,its dimension is reduced by the fast PCA algorithm,and then the dimensionality reduction data are normalized.Finally,the final classification and recognition of the experimental data is carried out by the Support Vector Machine（SVM）algorithm.Experimental results show that compared with single support vector machine,principal component analysis and directional gradient histogram algorithm,the proposed method can effectively improve the accuracy of face recognition under non-restrictive conditions and has a shorter recognition time.
作者
林椹尠
李相宇
惠小强
LIN Zhenxian1, LI Xiangyu2, XI Xiaoqiang3(1. School of Science, Xi＇an University of Posts and Telecommunications, Xi＇an 710121, China; 2. School of Telecommunications and Infoormation Engineering,Xi＇an University of Posts and Telecommunijcations, Xi＇an 710121,China;3 Instiute of Internet of Things and IT-based Industrialization,Xi＇an University of Posts and Telecommunications,Xi＇an 710061,China)
出处
《西安邮电大学学报》
2018年第2期49-54,65,共7页
Journal of Xi’an University of Posts and Telecommunications
基金
国家自然科学基金资助项目(61402367)
关键词
人脸识别
Haar特征分类器
梯度方向直方图特征
快速主成分分析
支持向量机
face recognition
Haar feature classifier
Histogram of Oriented Gradient （HOG） feature extraction
fast Principal Component Analysis （fast PCA）
Support Vector Machine （SVM）
