基于GVF Snake模型的人脸轮廓提取  

Facial contour extraction based on GVF Snake model

作　　者：宋书中[1] 高春庚[1] 马建伟[1] 

机构地区：[1]河南科技大学电子信息工程学院,河南洛阳471003

出　　处：《计算机工程与应用》2010年第14期209-211,共3页Computer Engineering and Applications

基　　金：国家自然科学基金No.70671035;河南省重点科技攻关项目No.082102210015~~

摘　　要：活动轮廓模型是计算机视觉领域的重要研究方向。针对传统的活动轮廓模型(Snake模型)对凹形轮廓处理效果差、初始轮廓必须充分接近图像边缘的缺点,通过改进外部能量项,提出了一种基于梯度矢量流活动轮廓模型的人脸轮廓提取算法。该算法把梯度矢量场作为外部能量场,克服了传统Snake模型力场范围小以及不能收敛于凹形边缘的缺点。实验结果表明,该方法能够快速、准确地提取人脸轮廓。Active contour model is an important research domain in computer vision.Aiming at the limitations that the traditional Snake model cannot converge to concave edge and the original contour must be close to the real edge of the image,a new facial contour extraction method based on Gradient Vector Flow（GVF） Snake model is proposed by improving the external energy.The algorithm uses GVF as its external energy.It not only expands the effective range of the energy field but also enhances the ability to deal with the concave edge.Experiments show that it can extract facial contour fast and exactly.

关 键 词：人脸轮廓提取 SNAKE模型 梯度矢量流 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]