摘要
本文提出一种新的基于特征点检测的参数化方法,针对不同面部表情,在三维面部模型存在丢失数据的情况下,改善面部识别准确度。然后采用混合插值方法,对FRGC(Face Recognition Grand Challenge,人脸识别大赛)数据库中的4950幅人脸图像进行了人脸特征点实验。Iterative Closest Point定位结果和特征点位置的估计数据证实了该方法的有效性。
This paper proposes a new parameterization method based on feature point detection to improve the accuracy of facial recognition in the case of missing data in the 3D facial model for different facial expressions.The main content of the research is to study the process of automatic embedding in the 3D face recognition system,any frontal face can be detected using traditional face segmentation and surface curvature information.Then,a hybrid interpolation method is used to conduct face feature point experiments on 4950 face images in the Face Recognition Grand Challenge(FRGC)database.In the Point Distribution Model(PDM),the unique facial features of the nose tip and the two corners of the eyes are used as the statistical input of the Iterative Closest Point(ICP).Finally,the percentage deviation of the average 3D profile is used to detect performance or feature point positioning.The positioning results and the estimated data of the location of the feature points confirmed the effectiveness of the proposed method.
作者
李骞
史岳鹏
彭勃
Li Qian;Shi Yuepeng;Peng Bo(School of Energy and Intelligence Engineering,Henan University of Animal Husbandry and Economy,Zhengzhou 450000)
出处
《中阿科技论坛（中英文）》
2021年第7期83-87,共5页
China-Arab States Science and Technology Forum
基金
2020年河南牧业经济学院博士启动资金项目“多模态农业大数据时空分析方法研究”(2020HNUAHEDF008)。
