标准正面人脸图象的特征提取

人脸的识别技术（ＦＲＴ）是当前模式识别领域的一个热点课题，人脸特征的自动提取是人脸自动识别过程中至关重要的一步，文中采用基于人脸几何特征的方法，设计一个初步的自动人脸特征提取系统，首先通过改进后的边缘检测和阈值技术在头肩型图象中找到头部轮廓，再利用“三停五眼”的标准进一步确定五官大概的位置，最后提取出７个有效的特征点，本系统建了一个包含５０个人脸图象的数据库，实验结果表明这种方法可以有效地获取头部轮廓一人脸特征点。

Extraction of the Features of Human Face Image in Standard Front-View

Face recognition technology (FRT) is a hot spot in the field of pattern recognition today. Automatic extraction of face image is the crucial step during automatic face recognition. By adopting the method based on geometric features of human face, an initial automatic system is designed for face feature extraction. At first, to find head outline in head shoulder image by improved edge detection and threshold technology. And then, to further define rough position of ears, nose and eyes by the norm of partitioning human face image into 3 parts vertically and 5 parts horizontally. And finally, to extract 7 effective feature points. A database involving 50 images of human face is built up by this system. Head outline and feature points of human face can be effectively achieved by this method, as shown by experimental results.