基于改进核判别算法的视频运动人脸图像相似表情识别研究  

Research on similar expression recognition of video moving face image based on improved kernel discriminant algorithm

作　　者：陈宇斌[1] CHEN Yu-bin(Zhangzhou Health Vocational College,Fujian Zhangzhou 363000,China)

机构地区：[1]漳州卫生职业学院,福建漳州363000

出　　处：《齐齐哈尔大学学报（自然科学版）》2021年第1期36-40,46,共6页Journal of Qiqihar University(Natural Science Edition)

基　　金：课程诊断与改进指标体系及实证研究(福建省中青年教师教育科研项目(社科类)(JAS19663))。

摘　　要：传统的视频运动人脸图像相似表情识别方法,利用小波变换提取人脸特征,后续特征分解效果不理想,导致表情识别准确度较低。为此,提出基于改进核判别算法的视频运动人脸图像相似表情识别研究。采用积分图法提取视频中的人脸表情,得到特征矩形区域。结合核判别算法对特征矩形区域进行分解得出表情特征矢量。结合弹性模板匹配法,计算和匹配表情特征矢量,得出匹配最优的表情,完成人脸图像相似表情识别。为验证所提方法的应用性能,设计仿真实验。实验结果表明,与传统方法相比,所提方法的表情识别准确度更高。本文设计识别方法具有应用有效性,为相关领域提供可靠依据。The traditional similar expression recognition method of video moving face image uses wavelet transform to extract face features.The subsequent feature decomposition effect is not ideal,resulting in low accuracy of expression recognition.In order to solve this problem,an improved kernel discriminant algorithm is proposed to recognize similar facial expressions in video moving face images.The integral graph method is used to extract the facial expression in the video,and the feature rectangle area is obtained.Combined with kernel discriminant algorithm,the feature rectangle region is decomposed to get the expression feature vector.Combined with the elastic template matching method,the expression feature vector is calculated and matched,and the optimal expression is obtained,and the similar expression recognition of face image is completed.In order to verify the application performance of the proposed method,simulation experiments are designed.Experimental results show that,compared with the traditional methods,the proposed method has higher accuracy in expression recognition.The identification method designed in this paper has application effectiveness and provides reliable basis for related fields.

关 键 词：人脸表情 核判别 特征矩形 特征矢量 表情类型 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]