【作者】 高淑欣； 穆志纯； 袁立； 徐晓娜；

【Author】 GAO Shu-xin MU Zhi-chun YUAN Li XU Xiao-na (University of Science & Technology Beijing,Beijing 100083)

【机构】 北京科技大学信息工程学院；

【摘要】 人耳识别技术已经成为一种新的生物特征识别技术。人耳图像的归一化在人耳识别中具有相当重要的意义,为后续的工作提供了前提。但传统的归一化方法往往是通过手动实现的。本文提出了一种新的基于改进的主动形状模型(ASM,Active Shape Model)的自动归一化技术。该算法通过对训练集中的人耳特征点建立点分布模型,然后在测试图像中搜索出外耳轮廓,最后以外耳长轴为基准将人耳图像归一到标准的尺寸和方向。更多还原

【Abstract】 Research of ear recognition has become a new technique of biometrics recognition.Normalization of ear image has a very important significance in ear recognition,and provides precondition for later work.Traditional method of normalization is always implemented manually.This paper proposed a new automatic normalization method based on improved Active Shape Model(ASM,Active Shape Model).This arithmetic creates Point Distributed Model using landmark points on ear in the training set,and then search out the outer contour of ear in the testing set.At last,normalize the ear image to normal size and direction according to the long axis of outer ear.更多还原