D0:10.13374h.issn1001-053x2013.11.020 第35卷第11期 北京科技大学学报 Vol.35 No.11 2013年11月 Journal of University of Science and Technology Beijing Nov.2013 基于二维和三维信息融合的人耳识别 李阳，穆志纯区 北京科技大学白动化学院，北京100083 ☒通信作者，E-mail:mu@ies.ustb.edu.cn 摘要针对人耳识别中存在姿态、光照变化等问题，提出信息融合的方法，将二维人耳和三维人耳的信息进行融合， 以克服姿态、光照对人耳识别的影响.对于二维人耳，由于姿态等的变化会导致人耳图像数据在高维空间中呈现出非线 性流形结构，采用等距映射这种流形学习算法进行特征提取，对三维深度人耳则采用3D局部二值模式进行特征提取， 然后分别进行二维和三维人耳识别，最后在决策层进行融合识别.在79人的人耳数据库上进行了实验，每人8幅带姿 态的二维人耳图像和6幅带光照的三维人耳深度图像实验结果表明，与单独的二维人耳和三维人耳识别相比，融合之 后的识别效果和认证效果均有很大的改善 关键词模式识别：信息融合：二维：三维：映射：人耳 分类号TP391.4 2D and 3D information fusion based ear recognition LI Yang,MU Zhi--chun☒ School of Automation and Electrical Engineering,University of Science and Technology Beijing,Beijing 100083,China XCorresponding author,E-mail:mu@ies.ustb.edu.cn ABSTRACT In order to solve pose and illumination variation problems in ear recognition,an information fusion method was proposed to fuse 2D and 3D ear information at the decision level.For a 2D ear,the ear images will become nonlinear manifold structure due to pose variation,so the manifold learning method,isometric mapping(Isomap),was used to extract features.For a 3D ear,the 3D local binary pattern(3DLBP)method was adopted for feature extraction. Then 2D ear recognition and 3D ear recognition were implemented separately.Finally,results from the 2D and 3D were fused at the decision level.Experiments were done on a database of 79 persons,one of which has eight 2D ears with pose variation and six 3D ears with illumination variation.It is found that both the recognition rate and verification rate are significantly improved compared with 2D ear recognition and 3D ear recognition alone. KEY WORDS pattern recognition;information fusion;two dimensional;three dimensional;mapping:ears 人耳识别技术是一种有效的生物特征识别方 的生理位置，尤其是在只能获得人脸侧面图像的情 法，具有良好的应用前景，受到了国内外众多研究 况下进行远距离身份识别时，人耳显得尤为重要 者越来越多的关注口.与人脸相比，人耳具有稳定 日前，国内外关于人耳识别技术的研究越来越 的结构特征，这些特征在人生的相当长的一段时间 多，但是大多是基于二维图像的，研究者提出了很 内都保持不变，不会随着年龄的变化而改变，而且 多用于二维人耳识别的方法【A-).文献[4提出 人耳不会受到表情变化所带来的影响四：与虹膜和 了基于小波变换的人耳识别方法，文献⑤提出了 指纹相比，人耳图像的获取是在非打扰的方式下进 基于力场转换的人耳识别方法，文献[6]提出了基 行的2-).此外，人耳具有丰富的结构特征和独特 于独立分量分析的人耳识别方法.这些研究结果表 收稿日期：2012-11-15 基金项目：国家自然科学基金资助项目(60973064)：北京市自然基金资助项目(4102039)：北京市重点学科资助项目(Xk100080537): 教育部博士点基金资助项目(20100006110014)1 35 ò 1 11 Ï  ® ‰ E Œ Æ Æ  Vol. 35 No. 11 2013 11  Journal of University of Science and Technology Beijing Nov. 2013 Äu‘Ún‘&EKÜ<£O o §;X ®‰EŒÆgÄzƧ® 100083 Ï&Šö§E-mail: mu@ies.ustb.edu.cn Á ‡ é<£O¥3^!1ìCz¯K§JÑ&EKܐ{§ò‘<Ún‘<&E?1Kܧ ±ŽÑ^!1ìé<£OKǑ. éu‘<§du^Cz¬<ãêâ3p‘m¥¥yњ‚ 56/(§æ^åNù«6/ÆSŽ{?1AÆJ§én‘Ý<Kæ^ 3D ÛÜŠª?1AÆJ§ ,￾©O?1‘Ún‘<£O§￾3ûü￾?1KÜ£O. 3 79 <<êâ¥þ?1 ¢§z< 8 ̑^ ‘<ãÚ 6 ̑1ìn‘<Ýã. ¢(JL²§†üÕ‘<Ún‘<£Oƒ'§K܃ ￾£OJÚyJþkéŒUõ. '… ª£O¶&EKܶ‘¶n‘¶N¶< ©aÒ TP391.4 2D and 3D information fusion based ear recognition LI Yang, MU Zhi-chun School of Automation and Electrical Engineering, University of Science and Technology Beijing, Beijing 100083, China Corresponding author, E-mail: mu@ies.ustb.edu.cn ABSTRACT In order to solve pose and illumination variation problems in ear recognition, an information fusion method was proposed to fuse 2D and 3D ear information at the decision level. For a 2D ear, the ear images will become nonlinear manifold structure due to pose variation, so the manifold learning method, isometric mapping (Isomap), was used to extract features. For a 3D ear, the 3D local binary pattern (3DLBP) method was adopted for feature extraction. Then 2D ear recognition and 3D ear recognition were implemented separately. Finally, results from the 2D and 3D were fused at the decision level. Experiments were done on a database of 79 persons, one of which has eight 2D ears with pose variation and six 3D ears with illumination variation. It is found that both the recognition rate and verification rate are significantly improved compared with 2D ear recognition and 3D ear recognition alone. KEY WORDS pattern recognition; information fusion; two dimensional; three dimensional; mapping; ears <£OEâ´«k)ÔAÆ£O {§äkûA^ µ§É IS ¯õïÄ ö5õ'5 [1] . †<òƒ'§<äk­½ (AƧù AÆ3<)ƒãžm SѱØC§Ø¬‘X #Cz UC§ … <جÉLœCz¤‘5KǑ [1]¶†Ú «ƒ'§<ã¼´3š‹6ªe? 1 [2−3] . d §<äk´L(AÆÚÕA )n § Ù´3U¼<òý¡ãœ ¹e?1ål°£Ož§<w Ǒ­‡. 8 §IS 'u<£OEâïÄ5 õ§´Œõ´Äu‘ã§ïÄöJÑ é õ^u‘<£O{ [4−6] . ©z [4] JÑ ÄuÅC†<£O{§©z [5] JÑ Äuå|=†<£O{§©z [6] JÑ Ä uÕá©þ©Û<£O{. ù ïÄ(JL ÂvFϵ2012-11-15 Ä7‘8µI[g,‰ÆÄ7℄ϑ8 (60973064); ®½g,Ä7℄ϑ8 (4102039); ®½­:Ɖ℄ϑ8 (Xk100080537); ÜƬ:Ä7℄ϑ8 (20100006110014) DOI:10.13374/j.issn1001-053x.2013.11.020