基于眼睛特征的人脸检测方法*

由于眼球的灰度级较低，而眼球周围白色区域的灰度级较高，因此在其交界处灰度级产生强烈突变。利用这一特征先从图像中找出可能的眼睛对，定位可能的人脸区域，通过计算其对称性来确定各种人脸特征的存在，更进一步验证可能的人脸区域。实验证明，此方法能迅速准确地从复杂背景图像中检测出人脸，而且对多人脸图像同样有效。

A Method of Face Detection Based on Eyes Feature

From the comparison of bright, it is easy to find that the gray level is lower in eyeball than the gray level of in white region around the eyeball. The gray level varies dramatically on the boundary between them. In order to get the feather, the first step is to find a pair of eyes in an image utilizing the feature of eye and locates possible face region. Then their symmetries can be calculated and the face can be verified from the different existing facial features. At last, the experimental results prove that this method can detect face in an image with complex background rapidly and accurately. Furthermore, it can perform well for multiple faces image.