作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2012, Vol. 38 ›› Issue (06): 184-186. doi: 10.3969/j.issn.1000-3428.2012.06.060

• 人工智能及识别技术 • 上一篇    下一篇

基于Contourlet变换与LPP的表情识别

高现文,付 炜,祝 鹏   

  1. (燕山大学信息科学与工程学院,河北 秦皇岛 066004)
  • 收稿日期:2011-08-22 出版日期:2012-03-20 发布日期:2012-03-20
  • 作者简介:高现文(1986-),男,硕士研究生,主研方向:模式识别,图像处理;付 炜,教授;祝 鹏,硕士研究生

Expression Recognition Based on Contourlet Transform and Locality Preserving Projection

GAO Xian-wen, FU Wei, ZHU Peng   

  1. (College of Information Science and Engineering, Yanshan University, Qinhuangdao 066004, China)
  • Received:2011-08-22 Online:2012-03-20 Published:2012-03-20

摘要: 提出一种基于Contourlet变换与局部保持投影(LPP)的人脸表情识别方法。将人脸表情图像分割为左眼(包括眉毛)、右眼(包括眉毛)和嘴三部分,利用Contourlet变换对局部表情图像和原始图像进行处理,得到图像的低频分量和高频分量。结合局部表情图像的低频分量与原始图像的高频分量,采用LPP算法提取表情特征,并利用支持向量机进行分类。实验结果表明,该方法的识别率较高。

关键词: Contourlet变换, 高频分量, 低频分量, 局部保持投影, 表情识别

Abstract: This paper presents an expression recognition method based on Contourlet transform and Locality Preserving Projection(LPP). It divides the face expressions image into three parts: left eye(including their eyebrows), right eye(including their eyebrows) and mouth. By using local Contourlet transform to process the local expression image and original image, it gets the low-frequency components and high-frequency components of the image. Combining with the low-frequency components of local expression image and high-frequency components of the original image, it extracts feature by using LPP algorithm, after that it uses Support Vector Machine(SVM) as the classifier. Experimental results indicate that the method has high recognition rate.

Key words: Contourlet transform, high-frequency component, low-frequency component, Locality Preserving Projection(LPP), expression recognition

中图分类号: