| 摘要: |
| 提出了一种基于特征曲线的快速人面识别方法.该方法首先对不同姿态和大小的人面进行定位和归一化,然后在归一化后的人面图像上建立多种特征曲线.作为对人面特征的描述,最后利用傅里叶描绘子对特征曲线进行解析提取关键特征,得到人面的表征向量.使用此方法在1300幅人面图像上进行了实验,结果表明，此方法在速度和准确度方面都具有较好的性能,而且对有不同姿态和表情的人面的识别具有一定的鲁棒性. |
| 关键词: 自动人面识别，人面归一化，积分投影，特征曲线，傅里叶描绘子. |
| DOI： |
| 分类号: |
| 基金项目:本文研究得到浙江省公安厅资金(No.1997-7-15)资助. |
|
| Automatic Human Face Recognition Based on Eigencurves |
|
ZHANG Gao-qing,WANG Shen-kang,YE Mao-dong
|
| Abstract: |
| A fast method for face recognition based on eigencurves is presented in this paper.First all faces with different poses and sizes are located and normalized.Then several kinds of eigencurves are built to describe characteristics of faces.And finally,the dominant characteristics are extracted which are composed of representative vectors of faces through analyzing the eigencurves using Fourier describer.This method has been tested on 1300 facial images and shown a good performance of speed and accuracy,and the result also shows it is robust for various head poses and expressions. |
| Key words: Automatic face recognition,face normalization,integral projection,eigencurve,Fourier describer. |
