Nest层次型分类器的快速、精确的眼睛定位算法  

A Method to Locate Eye Based on Nest-structured Real Adaboost Classifier

作　　者：刘福军[1,2] 高鑫[1] 

机构地区：[1]中国科学院电子学研究所,北京100190 [2]中国科学院研究生院,北京100190

出　　处：《科学技术与工程》2010年第17期4186-4191,共6页Science Technology and Engineering

摘　　要：人脸特征点定位是人脸识别、表情识别等人脸处理技术的基础,脸部特征点定位,尤其是眼睛的定位性能很大程度上影响着算法的精度。首先利用积分图像的概念快速计算眼睛样本的Haar-like特征,然后训练了Real Adaboost分类器,为了提高训练速度采用了层次型结构的Real Adaboost分类器。相比于传统Cascade结构的层次型分类器,本文采用了基于Nest结构的层次型分类器算法。研究表明这种新的算法,定位速度更快,定位精确度也更高。The locations of feature points of face,especially the locations of eyes,are very important for face recognition,expression recognition,and so on. Based on idea of integral images we first calculated the Haar-like features of eye samples,and then with those samples we trained Real Adaboost classifier. To reduce train time,we adopted staged classifier. Compared with the traditional Cascade-Structured classifier,we adopted Nest-Structured classifier. Experiments show that this Nest-Structured classifier can locate eyes more fast and more precisely.

关 键 词：HAAR-LIKE特征 积分图像 REAL ADABOOST分类器 Cascade结构 Nest结构 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]