基于嘴部特征的聋哑人情感识别方法  

Emotion recognition of deaf-mutes based on mouth feature

作　　者：张爱华[1] 王绍珍[1] 陈晓雷[1] 

机构地区：[1]兰州理工大学电气工程与信息工程学院,甘肃兰州730050

出　　处：《兰州理工大学学报》2016年第1期81-85,共5页Journal of Lanzhou University of Technology

基　　金：国家自然科学基金(61302116)

摘　　要：针对聋哑人在手语交流过程中,手部经常会遮挡面部五官,无法使用针对健听人的基于五官特征的情感计算方法来识别聋哑人的情感状态这一难题,提出一种聋哑人情感识别方法.该方法首先利用肤色信息从聋哑人面部图像中提取出嘴部区域,然后对提取到的嘴部区域进行腐蚀与膨胀运算,接着计算出嘴部区域特征值,并与训练样本中不同情感类型所对应的嘴部区域特征值进行比较,从而确定出聋哑人情感.实验结果表明,本文方法能快速准确地进行聋哑人情感判断.Aimed at the difficult problem in sign-language communication between deaf-mutes that it has no way of using the facial feature-oriented emotion computing method for normal healthy people to recognize the emotional state of the deaf-mutes due to frequent covering of their face with their hand, an emotion recognition method for deaf-mute is presented. In this method, the skin color information is used first to extract the mouth region of the deaf-mute from his facial image, then the corrosive and expansive opera- tion on the mouth region extracted is carried out, and then the eigen value of the mouth region is computed and compared to that of corresponding emotion variety in training samples, so that the emotion of the deafmute is determined. It is shown by experiment that, by using the method in this article, the emotion recognition of deaf-mutes can be performed rapidly and accurately.

关 键 词：手语视频 嘴部特征 特征矢量 情感计算 

分 类 号：TN911.72[电子电信—通信与信息系统]