计算机工程与应用 ›› 2011, Vol. 47 ›› Issue (12): 5-6.

• 博士论坛 • 上一篇    下一篇

多尺度变换域内混合投影熵的人脸特征描述

黄源源,李建平   

  1. 电子科技大学 计算机科学与工程学院,成都 611731
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2011-04-21 发布日期:2011-04-21

Facial feature descriptor using hybrid projection entropy in multi-scale transform domain

HUANG Yuanyuan,LI Jianping   

  1. School of Computer Science and Engineering,UESTC,Chengdu 611731,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2011-04-21 Published:2011-04-21

摘要: 提出一种新的人脸特征描述方法。使用Contourlet变换提取人脸图像低频子带,并对子带图像适当分块从而减少图像局部扭曲对识别的影响,利用混合投影函数和图像熵提取特征从而构建混合投影特征矩阵。在ORL、Yale、CMU PIE人脸数据库的实验表明该方法具有一定的优势。

关键词: 人脸识别, 轮廓波变换, 混合投影熵

Abstract: This paper proposes a new facial image feature description method.This method uses the Contourlet transform to get the low frequency sub-band and divides the sub-band image into several appropriate non-overlapping blocks so that the local image distortion will less affect the recognition result.It uses the hybrid projection function and image entropy to extract the features and construct the hybrid projection feature vector.The experimental results on ORL,Yale and CMU PIE face databases demonstrate that the new method is competitive.

Key words: face recognition, Contourlet transform, hybrid projection entropy