人脸图象的校准与特征抽取

AUTOMATIC NORMALIZATION AND FEATURE EXTRACTION OF FACE IMAGES

人脸自动识别是计算机模式识别领域的一个活跃课题 .文中提出了一个基于眼睛位置估计的人脸校准方法 ,并提出了一个基于 KL变换与具有统计不相关性的最佳鉴别变换抽取人脸有效鉴别特征的方法 .对由 96人、每人 10幅图象所组成的人脸图象数据库所作的实验结果表明 ,文中所提出的人脸校准方法能从具有一定复杂背景的人脸图象中比较成功地实现人脸自动校准 ,成功率与可靠性分别可达到 91.5 %与 99.5 % ,文中所提出的人脸特征抽取方法所抽取的人脸鉴别特征可得到接近 99%的正确识别率 .

The automatic recognition of human faces is an a ctive problem in the area of computer pattern recognition. This paper presents a new method to realize automatic normalization of face images based on estimatin g positions of two eyes, and presents a new method to extract effectively discri minant feature of face images. Experiments have been performed on a face databas e of 96 persons with 10 images per person. The experimental results show that th e new normalization method can successfully realize the automatic face normaliza tion to face images in a complex background with success rate and reliability as high as 91.5% and 99.5% respectively, and the correct recognition rate is near 99% on the discriminat feature of face images extracted by the new method.