Computer Engineering and Applications ›› 2010, Vol. 46 ›› Issue (35): 192-195.DOI: 10.3778/j.issn.1002-8331.2010.35.055

• 图形、图像、模式识别 • Previous Articles     Next Articles

Ear biometrics using local binary pattern and filter fusion

FENG Jun1,MU Zhi-chun2,MAO Wan-dui1   

  1. 1.School of Information Science and Technology,Shijiazhuang Tiedao University,Shijiazhuang 050043,China
    2.School of Information Engineering,University of Science and Technology Beijing,Beijing 100083,China
  • Received:2010-04-15 Revised:2010-08-23 Online:2010-12-11 Published:2010-12-11
  • Contact: FENG Jun

基于局部二值模式与滤波器融合的人耳识别

封 筠1,穆志纯2,毛晚堆1   

  1. 1.石家庄铁道大学 信息科学与技术学院,石家庄 050043
    2.北京科技大学 信息工程学院,北京 100083
  • 通讯作者: 封 筠

Abstract: An important approach for ear biometrics is provided by rich texture information of ear image.This paper builds two kinds of novel texture descriptor by the fusion of two-dimension filter transformation and Local Binary Pattern(LBP) descriptor,i.e.local Gabor binary pattern descriptor,and local wavelet binary pattern descriptor.An ear recognition scheme is proposed based on the local texture feature descriptors.The experimental results for USTB ear image set 3 show that 98.14% cross-validation recognition rate is obtained by 1-NN classifier under Chi square similarity measure and local Gabor binary pattern descriptor with less run time.

Key words: ear biometrics, Local Binary Pattern(LBP), Gabor filter, wavelet transform, K nearest neighbor

摘要: 人耳的丰富纹理信息为身份鉴别提供了重要解决途径。融合信号滤波变换与局部二值模式算子,构建了两种新型纹理描述算子,即局部Gabor二值模式算子与局部小波二值模式算子;提出了一套基于局部纹理特征描述算子的人耳识别方案。针对USTB人耳图像库三的测试实验表明,在选用Chi方相似性测度和局部Gabor二值模式算子时,1-NN分类器可获得98.14%的交叉验证识别率,且运行时间较少。

关键词: 人耳识别, 局部二值模式, Gabor滤波, 小波变换, K近邻

CLC Number: