许良凤,王家勇,胡敏,林辉,侯登永,崔婧楠.局部梯度DT-CWT主方向模式的表情识别[J].电子测量与仪器学报,2017,31(4):522-529
局部梯度DT-CWT主方向模式的表情识别
Facial expression recognition based on local gradient DT-CWT dominant direction pattern
  
DOI:10.13382/j.jemi.2017.04.005
中文关键词:  表情识别  双数复小波变换  主方向模式  梯度方向  分块直方图
英文关键词:expression recognition  dual tree complex wavelet transform  dominant direction pattern  gradient direction  block histogram
基金项目:国家自然科学青年基金(61300119)、国家自然科学基金重点项目(61432004)、安徽省自然科学基金(1408085MKL16)资助项目
作者单位
许良凤 合肥工业大学 情感计算与先进智能机器安徽省重点实验室合肥230009 
王家勇 合肥工业大学 情感计算与先进智能机器安徽省重点实验室合肥230009 
胡敏 合肥工业大学 情感计算与先进智能机器安徽省重点实验室合肥230009 
林辉 合肥工业大学 电子科学与应用物理学院合肥230009 
侯登永 合肥工业大学 情感计算与先进智能机器安徽省重点实验室合肥230009 
崔婧楠 华南理工大学广州510006 
AuthorInstitution
Xu Liangfeng Anhui Province Key Laboratory of Affective Computing and Advanced Intelligent Machine, Hefei University of Technology, Hefei 230009, China 
Wang Jiayong Anhui Province Key Laboratory of Affective Computing and Advanced Intelligent Machine, Hefei University of Technology, Hefei 230009, China 
Hu Min Anhui Province Key Laboratory of Affective Computing and Advanced Intelligent Machine, Hefei University of Technology, Hefei 230009, China 
Lin Hui School of Electronic Science and Applied Physics, Hefei University of Technology, Hefei 230009, China 
Hou Dengyong Anhui Province Key Laboratory of Affective Computing and Advanced Intelligent Machine, Hefei University of Technology, Hefei 230009, China 
Cui Jingnan South China University of Technology, Guangzhou 510006, China 
摘要点击次数: 351
全文下载次数: 276
中文摘要:
      提出一种采用局部梯度双树复小波变换(dual tree complex wavelet transform, DT CWT)主方向模式(dominant direction pattern, DDP)的人脸表情识别方法。首先,对归一化后的表情图像进行4层DT CWT,每一层得到8个方向的DT CWT特征图像,其中包括6个高频方向和2个低频方向,构建一种新的主方向模式(IDDP)对每个DT CWT特征图像进行编码;然后按照基于梯度方向的融合规则将每一层上IDDP编码特征图融合到一起,将融合图进一步划分为若干个不重叠且大小相等的子块,分别计算每个子块区域的直方图分布,将其联合起来得到人脸表情图像的特征;最后,采用基于Fisher加权的Chi平方概率统计最近邻方法进行分类识别。大量实验表明,算法在识别率和识别时间上都体现出了一定的优势。
英文摘要:
      A novel facial expression recognition is proposed in the paper, in which the local gradient dual tree complex wavelet transform dominant direction pattern is used. Firstly, four layers DT CWT are used on normalized expression image. For each layer, we can obtain the feature images of eight directions, which include 6 high frequency directions and 2 low frequency directions. A new DDP (IDDP) is constructed, and which is used to code for each DT CWT feature image. Secondly, the IDDP feature images of each layer in different directions are fused based on rules of gradient direction, and every fused image is divided into several non overlapping and equal sized blocks. The corresponding histogram of the fused feature in each block is calculated respectively, and the final feature of facial expression image is obtained by cascading all of them. Finally, the nearest neighbor method based on Chi Square statistic weighted by Fisher is used to classify and identify. A large number of experiments show that the proposed method has a certain advantage on the recognition rate and recognition time.
查看全文  查看/发表评论  下载PDF阅读器