【摘要】 利用球面谐波模型建立的9D光照子空间进行人脸识别。用SFS方法获得人脸表面特征点深度,以特征点深度为基础,用NURBS方法重构三维人脸,并用该三维人脸的法向量计算球面谐波基图像,进行人脸识别。结果表明,以所求的球面谐波基图像为基础进行人脸识别,不仅提高了人脸表面法向量的精度,减少了运算量,而且使得人脸的识别率明显提高。利用球面谐波模型进行人脸识别不仅简单而且精度也较高。更多还原

【Abstract】 This paper uses a spherical harmonic model to create 9-D illumination subspace in face recognition.It makes use of SFS method to obtain the depth of facial surface feature points.On this basis,it uses NURBS method to achieve three dimensional facial reconstruction,and the new method to obtain the normal vector of three dimensional face.Then it uses the normal vector to calculate the spherical harmonics basic images and to take face recognition.The experimental results show that the new method based on the spherical harmonics basic image not only improves the face surface normal vector accuracy and reduces computation,but also improves the face recognition rate significantly.更多还原