摘要
针对以往利用人脸图像单方面进行性别识别或年龄估计,提出了利用公共特征、私有特征同时进行性别识别与年龄估计。用对光照、尺度变化具有很强鲁棒性的Gabor小波变换提取人脸特征。降维后的有效人脸特征分成公共特征、私有特征两部分,公共特征用于性别识别,私有特征进行年龄估计。在FG-NET人脸库及自建OFID人脸库中用RBF神经网络进行了实验,取得了良好效果。
Compared to the one-sidedness of gender classification or age estimation based on facial image in the past,a novel method based on public features and private features for gender classification and age estimation is proposed.Face features are extracted by Gabor wavelet transform which are robust to the illumination change and scale variations.Effective face features which have been reduced dimension are divided into public features and private features,public features are used for gender classification,private features are used for age estimation.The experimentation is conducted with radial basis function neural network in FG-NET face database and own OFID face database,and very promising results are achieved.
出处
《计算机工程与设计》
CSCD
北大核心
2012年第5期1997-2001,共5页
Computer Engineering and Design
基金
山西省教育科学"十一五"规划课题基金项目(GH09223)
关键词
特征提取
公共特征
私有特征
性别识别
年龄估计
feature extraction
public feature
private feature
gender classification
age estimation
