作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2011, Vol. 37 ›› Issue (9): 1-3. doi: 10.3969/j.issn.1000-3428.2011.09.001

• 博士论文 •    下一篇

基于三维模型的人脸姿态估计方法

曾 慧,穆志纯,袁 立   

  1. (北京科技大学信息工程学院,北京 100083)
  • 出版日期:2011-05-05 发布日期:2011-05-12
  • 作者简介:曾 慧(1980-),女,讲师、博士,主研方向:三维计算机视觉,模式识别;穆志纯,教授、博士生导师;袁 立,讲师、博士
  • 基金资助:
    国家自然科学基金资助项目(61005009, 60973064);北京市自然科学基金资助项目(4102039)

Face Pose Estimation Method Based on 3D Model

ZENG Hui, MU Zhi-chun, YUAN Li   

  1. (School of Information Engineering, University of Science and Technology Beijing, Beijing 100083, China)
  • Online:2011-05-05 Published:2011-05-12

摘要: 提出一种三维人脸姿态估计方法。该方法通过估计三维平面人脸模型到图像平面的单应矩阵来获得人脸相对于摄像机坐标系的旋转矩阵,并利用M-估计优化方法迭代求精。其主要特点是:实施简单,不需要对透视摄像机参数预先进行标定,能够在较大范围内较精确地估计人脸姿态。对模拟数据及真实人脸图像的实验均验证了该方法的有效性。

关键词: 摄像机标定, 三维模型, 单应矩阵, M-估计, 人脸姿态估计

Abstract: A method is proposed for 3D face pose estimation. The face pose are determined by estimating the homography matrix between the face model plane and the image plane. The M-estimator is employed to further refine the results. The main characteristic of the method is that it is simple to implement, does not need to calibrate the projective camera beforehand and can recover the face pose satisfactorily under a wide degree of head motion. Both simulation and real data results show it is effective.

Key words: camera calibration, 3D model, homography matrix, M-estimation, face pose estimation

中图分类号: