基于差分能量图和CGBP的微表情识别

发布时间：2017-12-31 20:13

本文关键词：基于差分能量图和CGBP的微表情识别　出处：《西南交通大学学报》2016年06期 　论文类型：期刊论文

【摘要】：微表情是一种不能自主控制和伪装的面部表情,其与诚信度的关系密切,具有持续时间短且难以识别的特征.为提高计算机自动识别微表情的准确性,提出一种基于差分能量图和中心化Gabor二值模式(centralized Gabor binary patterns,CGBP)的微表情识别方法.该方法首先利用差分法计算微表情序列的能量得到差分能量图,获得人脸面部肌肉相位的变化;其次将Gabor与中心二值模式CBP相结合,得到CGBP算子对能量图进行微表情的特征提取;最后利用ELM分类器进行微表情分类识别.在CASME微表情库上的实验结果表明,该方法比LBP-TOP、DTSA3、Gabor、VLBP、CBP-TOP算法更能有效地获得微表情序列的时空纹理特征,平均识别率为86.54%.
[Abstract]:Microfacial expression is a kind of facial expression which can not be controlled and camouflaged independently. It is closely related to the degree of honesty and has the characteristics of short duration and difficult to recognize. In order to improve the accuracy of automatic recognition of microexpressions by computer. A centralized Gabor binary patterns based on differential energy diagram and central Gabor binary mode is proposed. In this method, the difference method is used to calculate the energy of microfacial expression sequence to obtain the difference energy map, and the phase change of facial muscle is obtained. Secondly, the Gabor is combined with the central binary mode CBP, and the CGBP operator is obtained to extract the feature of the micro-expression of the energy map. Finally, the ELM classifier is used to classify and recognize microexpressions. The experimental results on CASME microfacial expression database show that the proposed method is better than LBP-Topo DTSA3 / GaborBP. The CBP-TOP algorithm is more effective in obtaining temporal and spatial texture features of microfacial expression sequences with an average recognition rate of 86.54.
【作者单位】：河北工业大学计算机科学与软件学院;中国汽车技术研究中心;
【基金】：国家自然科学基金资助项目(60302018) 天津市科技计划资助项目(14RCGFGX00846) 河北省自然科学基金资助项目(F2015202239)
【分类号】：TP391.41
【正文快照】： 人脸表情识别的目的是在非接触的情况下通过计算机自动识别出人脸的表情,进一步分析出人的情感或情绪,是智能化人机接口领域的一个重要研究课题[1-2].然而,人除了有普通表情,还存在着两种难以被人觉察的表情,一种是强度非常低的弱表情(subtle expression),另一种是持续时间非

【相似文献】