欧宗瑛

个人信息Personal Information

教授

博士生导师

硕士生导师

性别:男

毕业院校:大连工学院

所在单位:机械工程学院

电子邮箱:ouzyg@dlut.edu.cn

扫描关注

论文成果

当前位置: 中文主页 >> 科学研究 >> 论文成果

Eyes location by hierarchical SVM classifiers

点击次数:

论文类型:会议论文

发表时间:2004-08-19

收录刊物:EI

卷号:3173

页面范围:611-615

摘要:This paper presents a method for eyes location using a two-level hierarchy of SVM (Support Vector Machines) classifiers. On the first level, a two-eye region classifier is obtained by training the SVM using grayscale projections of the two-eye region images. Utilizing this classifier, the region where the two eyes lie can be located by searching the whole face image. On the second level, the left and right eye classifier are obtained by training SVM using grayscale of left and right eye images respectively. Using these two classifiers, the two eyes can be precisely located by searching the output region of the first level. Experimental results show that this method is sufficiently generic and can cope with more various image conditions than exiting techniques. © Springer-Verlag Berlin Heidelberg 2004.