基于普通样本的Adaboost人脸检测算法  

Face detection algorithm based on general sample Adaboost

作　　者：李丽娟[1] 张健[1] 

机构地区：[1]湖南大学计算机与通信学院,长沙410082

出　　处：《微计算机信息》2010年第35期208-209,共2页Control & Automation

摘　　要：现实中,各个图像的检测难度不一,并且检测难度处于中间状态的图像占大多数。利用此规律,本文提出了一种改进的AdaBoost算法——普通样本AdaBoost。该方法首先分析了训练样本的检测难度分规律,并以此确定普通样本的检测率范围。在训练时,对于普通样本采用特殊的权重更新方法,而非普通样本则使用传统方法。实验结果表明,新算法比传统AdaBoost算法在检测率和负样本误检率上作的更好。In the real world,every picture has its own detection difficulty and most of the detection rates are moderate.This paper presented an advanced algorithm—general sample adaboost by taking use of this law which focused on the samples whose detection rates were moderate.First,the method defined the scope of general sample by analysising the law of the detection rate distribution of training sample.During training,the weight updating method for the general samples was specially designed,and the non-general samples use the general method.The experiment results show the new method gets better FNR and detection rate.

关 键 词：ADABOOST 人脸检测 权值更新 普通样本 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]