摘要
提出了一种基于多尺度Gabor特征的三维人脸识别方法.首先将预处理后的三维人脸模型映射至平面上的参数化网格,再利用线性差值得到空间三维网格的二维几何图像.然后利用多尺度Gabor变换将几何图像分解为不同尺度下包含不同频率、不同方向人脸信息的Gabor响应系数,并提取垂直低频Gabor响应作为人脸的Gabor特征.最后计算不同尺度下Gabor特征的相似度并融合为人脸识别的总相似度.在FRGC v2.0数据库中进行的大量实验表明,提出的方法识别效果较好,提取的人脸Gabor特征具有较好的身份表征性.
A 3D face recognition method based on multi-scale Gabor features is proposed.First,the preprocessed 3D face model is mapped into a planar parameterized mesh.A 2D geometry image of the spatial 3D mesh is obtained by means of linear interpolation.Then the geometry image is decom-posed into Gabor responses of different scales,frequencies and orientations,among which the verti-cal Gabor responses of low frequencies are extracted as the facial Gabor features.Finally,similarities of multi-scale Gabor features are computed and fused as an overall similarity score.Extensive experi-ments are conducted on the FRGC v2.0 database,and the results verify that the facial Gabor features extracted by the proposed method can effectively represent the identity.
出处
《东南大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2013年第6期1212-1216,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)
江苏省高校自然科学研究计划资助项目(13KJB510015)