摘要
提出一种几何活动轮廓模型初始轮廓自动分割的方法。此方法综合利用了分水岭算法以及几何活动轮廓模型算法,通过利用人脸的肤色特征、结构特征等逐级判别原理,从而实现了自动分割的目的。实验结果表明,采用此方法很好地实现了目标对象轮廓的自动分割,基本可达到与人工干预相近的效果。此方法同样可以广泛应用于人脸检测、图像自动分割、视频监控系统等领域。
A method,which adopts the watershed algorithm and geometric active contours model to accomplish the image segmentation, is proposed in this article. Via the step-by-step differentiation of the colors of the facial skin and structure characteristics of human beings, the purpose of decreasing the human interference in initial extracted face contour and realizing the automatic choosing in confirming people＇s face region is achieved. Experimental result shows this method do as well as the traditional method. This new method can be widely applied to the field as monitoring system.
出处
《计算机应用与软件》
CSCD
北大核心
2006年第12期8-9,22,共3页
Computer Applications and Software
基金
国家863计划项目2002AA145090资助。
关键词
人脸轮廓提取
几何活动轮廓模型
分水岭算法
Face contour extraction Geometric active contours Watershed algorithm
