基于3D模型的虚拟人脸生成算法和多姿态人脸识别
Virtual Face Generation Algorithm and Multi-Pose Face Recognition Based on 3D Model
本文介绍了一种基于3D模型的虚拟人脸识别算法,它利用有限姿态图像通过旋转公式生成任意角度的虚拟人脸图像,然后基于多姿态的测试图像和虚拟图像进行人脸识别。文中首先介绍了人脸偏转在3D空间与二维平面的关系,并且推导出二维平面中人脸偏转与偏转角度的函数关系式,然后利用该关系式生成多姿态虚拟图像,最后利用Fisherface方法进行人脸识别。实验结果表明基于3D模型的虚拟人脸生成算法的多姿态人脸识别有较高的正确率。
virtual face recognition algorithm based on 3D model is proposed in this paper. Multi-pose virtual face images are synthesized by formula and several different pose face images, and then the face recognition is completed based on multi-pose virtual face images. At first step, the relationship of face deflection in 3D space and two dimensional is introduced, and derive the relation function between face deflection and angle in two dimensional. And then, multi-pose virtual face images are synthesized by the relation function. Finally, we design the face recognition experiment based on Fisherface. The experiment results show that the performance of the algorithm discussed in the paper has significantly increased.
房斌、刘润
计算技术、计算机技术
多姿态虚拟人脸人脸识别3D模型二维空间
multi-posevirtual faceface recognition3D modeltwo-dimensional
房斌,刘润.基于3D模型的虚拟人脸生成算法和多姿态人脸识别[EB/OL].(2015-11-30)[2025-08-23].http://www.paper.edu.cn/releasepaper/content/201511-805.点此复制
评论