摘要
对于有背景的彩色图像,肤色是人体表面最显著的特征之一,所以肤色特征是人脸检测中一个重要的特征[1～2]。肤色特征主要由肤色模型描述,检测方法可以分为颜色选择,肤色区域分割和人脸检测三个步骤。文章提出的肤色模型可以较好的适应光照变化,采用肤色分割的方法,可以快速检测不同大小,不同平面以及一定侧面旋转角度的人脸。对简单背景下的人脸检测的检测率达到95.65%,复杂背景下的人脸检测的检测率达到85.22%。
For color images with certain background,color of skin is the most important character of the body surface,so skin color is an important character in face detection.Skin character can be described by skin color model.The detection method can be divided into color selection,skin color segmentation and face detection.The skin color model was presented can well adapt to different light condition.Using the skin color based face detection,different sizes,different plane and a certain rotation angle of the face could be quickly detected.In the simple context picture,the face detection rate reach to 95.65%,in the complex context picture,the face detection rate can reach to 85.22%.
出处
《计算机与数字工程》
2012年第3期102-103,108,共3页
Computer & Digital Engineering
关键词
人脸检测
肤色模型
肤色分割
face detection
skin color model
skin color segmentation
