J4

• 计算机科学 • 上一篇    下一篇

基于椭圆模型和神经网络的人脸姿态估计方法

刘淼, 郭东伟, 马捷, 孙浩翔, 周春光   

  1. 吉林大学 计算机科学与技术学院, 长春 130012
  • 收稿日期:2007-10-24 修回日期:1900-01-01 出版日期:2008-07-26 发布日期:2008-07-26
  • 通讯作者: 郭东伟

Face Pose Estimation Based on Elliptical Model and Neural Network

LIU Miao, GUO Dongwei, MA Jie, SUN Haoxiang, ZHOU Chunguang   

  1. College of Computer Science and Technology, Jilin University, Changchun 130012, China
  • Received:2007-10-24 Revised:1900-01-01 Online:2008-07-26 Published:2008-07-26
  • Contact: GUO Dongwei

摘要: 基于椭圆模型和神经网络方法, 提出一种人脸姿态估计方法. 先采用椭圆模型对人脸边界进行拟合, 确定眼睛和嘴在椭圆中的位置, 再应用BP神经网络对位置参数和人脸姿态参数的关系进行学习, 从而对人脸姿态进行估计. 实验结果表明, 用该方法对较大范围内的人脸转动进行估计, 结果比较准确.

关键词: 脸姿态估计, 椭圆模型, 人工神经网络

Abstract: The present paper presents a new method based on the elliptical model and artificial neural network for estimating the pose of human face. First, elliptical model is used to fit the boundary of face and the positions of eyes and mouth are located. Second, the BP neural network is trained to learn the relationship between position parameters of eyes and mouth and the parameters of pose. The experimental results demonstrate that this method can estimate accurately pose of faces rotating in a large scope.

Key words: face pose estimation, elliptical model, artificial neural network

中图分类号: 

  • P391.4