摘要
8位灰度图、24位彩色图的人脸识别,采用基于人脸亮度特征的人脸识别方法-亮度曲线匹配块数法。8位灰度图像的人脸识别步骤包含人脸图像质量判断及图像预处理、人脸图像特征提取及人脸识别模型建立。24位彩色图像的人脸识别关键步骤为肤色过滤、人脸定位,将YIQ和KL色系结合进行肤色过滤以实现人脸的定位。实验中全部图像以BMP格式存储,并采用130×90归一化图像,得到了较好的识别效果。
Aiming at the face recognition of 8-bit gray images and 24-bit color images, a simple, effctive method named brightness curve and matching number for face recognition based on the brightness feature of face is put forward. Aiming at the face recognition of 8-bit gray images, the step includes the quality estimation of face images and images pretreatment, the feature extraction of face images and the creation of face recognition model. Aiming at the face recognition of 24-bit color images, the step is skin-color filter and face location. YIQ color coordinate and KL color coordinate are combined to make a skin filter in order to make face location. In the experiment, all the images are stored as BMP format, and unitary images in the form of 130×90 are used, satisfactory effect is obtained.
出处
《兵工自动化》
2005年第5期79-81,共3页
Ordnance Industry Automation
关键词
人脸识别
亮度特征
肤色模型
人脸定位
Face recognition
Brightness feature
Skin-color model
Face location
