基于多摄像机的人脸三维建模  

3D Facial Modeling based on Several Cameras

作　　者：张立臣[1] 毛庆[1] 冯德民[1] 王映辉[1] 

机构地区：[1]陕西师范大学计算机学院,西安710062

出　　处：《电脑开发与应用》2005年第4期2-3,共2页Computer Development & Applications

基　　金：陕西省自然科学基金 (2003F35);陕西师范大学校级重点基金 (995147)资助。

摘　　要：为了提高三维人脸建模的精确度,在双目立体视觉理论的基础上,提出了一种基于多台摄像机拍摄的人脸三维建模方法。该方法利用两组摄像机分别在不同的角度拍摄人脸,并采用双目立体视觉理论计算三维人脸信息,最后结合两组摄像机分别得到的三维人脸信息,进行坐标变换并重构三维人脸。该方法有效弥补了原来由两台摄像机拍摄人脸所得的三维信息的精确度较差的缺点,并具有一定的实用性。In order to improve the precision of 3D facial modeling and according to binocular stereovision theory,a method of 3D facial modeling using several cameras is put forward.This method adopts two groups of cameras to shoot the face of the same person from different angles respectively,and then calculates 3D facial information based on binocular stereovision theory.In the end,integrating 3D facial information that is produced through two groups of cameras,the final 3D facial information is calculated through coordinate transformation.This method effectively compensates the lower precision of 3D facial information shot by two cameras,and also has some degree of practicability.

关 键 词：三维人脸建模 双目立体视觉 坐标变换 多摄像机 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]