基于肤色模型和椭圆环模板的人脸跟踪及姿态估计  

Face Tracking and Pose Estimation with Skin Color Model and Ellip tical Model

作　　者：马淑燕[1] 孔德慧[1] 尹宝才[1] 郭金铜[1] 

机构地区：[1]北京工业大学多媒体与智能软件技术实验室,北京100022

出　　处：《计算机工程与应用》2004年第17期50-55,105,共7页Computer Engineering and Applications

基　　金：国家973基础研究发展规划项目(编号:2001CCA03300);国家863高技术研究发展计划项目(编号:2002AA119010-14);北京市教委项目(编号:P070702-01);北京工业大学研究生科技基金(编号:YKJ2002-14)的资助

摘　　要：文章提出了一种基于肤色模型,结合椭圆环模板进行人脸跟踪及姿态估计的算法。该算法在基于肤色模型实现人脸跟踪及特征定位的过程中首先利用肤色模型定位人脸肤色区域,在跟踪中增加了自适应学习模块,使得原始的肤色模型能够在不同光照下实现自适应调整。然后利用人脸形状的先验知识,通过椭圆环模板实现人脸边缘的精确定位。最后根据所得到的面部特征和人脸边缘位置估计出人脸的姿态。实验表明,该算法能够在自然光照条件下取得较为满意的跟踪结果,同时对人脸在旋转、缩放、遮挡等条件下,多人脸背景下的跟踪有较强的鲁邦性。This paper presents an algorithm for facial tracking and pose estimation based on skin color model and el-liptical template.Firstly,the tracked face and the facial feature are located roughly using the skin color model.During face tracking the self-adjusting skin color model is used,which can be adjusted automatically in different illumination.Secondly,with the facial shape information,the elliptical template locates precisely the contour of the face.Finally,the face pose is estimated with the location of the facial features and the contour of the face.Experiential result shows that the algorithm has the satisfying tracking result under the natural circumstances.The most important thing is that the al-gorithm is very robust to rotation,tilting,brief occlusion and multiple moving people in the background.

关 键 词：肤色模型 椭圆环模板 特征定位 姿态估计 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]