计算机应用与软件2012，Vol.29Issue(12)：24-26,3.DOI:10.3969/j.issn.1000-386x.2012.12.007
基于不平衡分类的人脸检测系统
FACE DETECTION SYSTEM BASED ON IMBALANCED CLASSIFICATION
摘要
Abstract
Face detection is a key technology in biometric identification field. Face detection system based on BalanceCascade imbalance classification algorithm is proposed for the characteristic of class-imbalance of positive and negative samples in face recognition. The system makes the positive and negative samples in each layer have similar size by controlling the false alarm rate of classifier, then weights all the weak classifiers to construct final strong classifier to eliminate the class-imbalance characteristics of the positive and negative samples. The experiments in ORL face dataset use F-measure and AUC as the evaluation criteria. Compared with traditional imbalance classification algorithms AdaBoost and UnderSampling, the experimental results show that the BalanceCascade algorithm is superior to the traditional imbalance classification algorithms.关键词
人脸识别/不平衡分类/误报率Key words
Face detection/Imbalanced classification/Falsw alarm ratekeywords分类
信息技术与安全科学引用本文复制引用
孙玉,刘贵全,汪中..基于不平衡分类的人脸检测系统[J].计算机应用与软件,2012,29(12):24-26,3.基金项目
国家自然科学基金项目(61073110). （61073110）
