摘要
针对三维人脸识别,提出了一种基于面部等测地轮廓线并结合局部特征和整体特征的人脸识别方法.首先,在人脸中提取到鼻尖点等测地距离的点组成等测地轮廓线来表征人脸面部曲面;然后,根据重采样后轮廓线上点的邻域信息提取局部特征,根据轮廓线的整体形状信息提取人脸整体特征;最后,分别利用比较局部特征和整体特征,将比较结果在决策级融合,给出最终识别结果.所提算法在FRGC(face recognition grand challenge)v2.0数据库上进行测试,测试结果表明,特征融合后的识别性能优于单一特征的识别率,且具有较好的表情鲁棒性.
A 3D face recognition method combining local and global geometric features which are extracted from the iso-geodesic curves is proposed.First,a set of facial curves with different geodesic distances from the nose tip are extracted to represent a facial surface.Then,for each point in the re-sampled facial curves,local feature which is invariant to pose is calculated from its local neighborhood and the local feature represents the geometric information of the local neighborhood.Next,the shape information of the facial curves is computed which constitute the global feature.Finally,local feature and global feature are compared respectively,and the final result is the weighted sum of them.The method is tested on the FRGC（face recognition grand challenge） v2.0 data set,and the experimental results show that recognition performance using compositional features is superior to that using single feature.Furthermore,it is also robust to expression.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2012年第4期618-622,共5页
Journal of Southeast University：Natural Science Edition
基金
国家自然科学基金资助项目(51175081
61107001)
江苏省自然科学基金资助项目(BK2010058)
关键词
三维人脸识别
等测地轮廓线
特征融合
主成分分析
3D face recognition
iso-geodesic curves
feature fusion
principal component analysis
