基于局部二元模式算子的人脸性别分类方法

提出了两种基于局部二元模式(Local Binary Pattern,LBP)算子的人脸性别分类方法:级联LBP方法和boosting LBP方法.前一种方法遵循从局部到整体的级联策略,使用LBP算子对由小波分解得到的细节图像进行特征提取,以达到扩充特征提取范围和增强所提取特征的有效性的目的,随后采用自适应加权机制对人脸图像的各个分块赋以不同的权值.后一种方法采用可变尺寸的子窗口对人脸图像进行扫描,在扫描所得的每个子窗口中,使用LBP算子对该子窗口提取LBP直方图.计算样本图像的LBP直方图和模板的LBP直方图之间的度量,并由此构建弱分类器集.利用Adaboost算法选取最有效的若干个弱分类器集组合成为强分类器.进行了三个基于LBP算子的人脸性别分类实验,实验所使用的训练集和测试集皆选自FERET人脸数据库.实验结果证明:LBP算子能有效地从人脸图像中提取出针对人脸性别分类的特征,并可以达到人脸性别分类的目的.所提出的两种基于LBP算子的方法可以有效的解决传统LBP方法所存在的特征提取范围有限、加权机制客观性不足等问题.

Gender classification based on local binary pattern

In this paper,we present two novel approaches for gender classification by local binary pattern(LBP) based classifiers.The first one is cascade LBP method.In this method,we apply wavelet to decompose images into four kinds of frequency images.Then we extract LBP features with the local to holistic way to make features more representative.And,the adaptive weight mechanism is adopted to show the different importance of feature data.The second one is the boosting LBP method,in which the face area is scanned with scalable small windows from which LBP histograms are obtained to effectively express the local feature of a face image.The Chi square distance between corresponding LBP histograms of sample image and template is used to construct weak classifiers pool.Adaboost algorithm is applied to build the final strong classifier by selecting and combining the most useful weak classifiers.In addition,several experiments are made for classifying gender based on local binary pattern.The male and female image set is collected from FERET database.Finally,the results of experiments show that the features extracted by LBP operator are discriminative for gender classification and our proposed approaches achieve better performance of classification than several other methods.