（北方工業大學 信息工程學院，北京市 石景山區 100144）
摘 要:人耳作為一種新型的生物特征具有許多優點。相比于虹膜識別，人臉識別等，人耳具有可遠距離拍攝、大小、結構等在成年以后基本不發生變化的明顯優勢。目前，二維的人耳識別方法如果希望獲得較高的識別率，需要一些限制條件，比如姿態、光照、像素分辨率等；同時，頭發、耳飾等物件的遮擋也會對人耳識別的準確率產生較大影響。本文利用基于力場轉換算法(Force Field Transform)提取力場的散度特征(Divergence Features)，然后加入人耳矩特征，通過sift算法進行身份識別，最后在人耳圖像庫上進行測試，實驗結果表明該方法具有較高的識別率。
關鍵字：人耳識別 力場轉換 散度特征 SIFT
中圖分類號： TP391.41 文獻標識碼：A 文章編號：
Ear Divergence Feature Extraction and Matching Based on Force Field Transformation
CUI Yan-Wei
(North China University of Technology, Beijing 100144 China .Yan-Wei Cui, cuiyanwei1144@163.com)
Abstract: As a novel biometric feature, ear recognition has many advantages compared to other biometric technology like iris and face recognition. It has distinct advantages with its time-invariant attributes of size and structure in human's adult life, and the ear images can be easily captured from a long distance. At present, some restrictions are needed to be considered for better recognition result when using 2-D digital image data, like from different angles, under different lighting conditions, and with different cameras. The recognition accuracy must also be affected by the background clutter when ears are partly occluded by hair, earrings and other objects. In this paper, we extract ear features based on force field transform and obtain divergence features, combined with the method of ear moment, using SIFT algorithm for the identification of individuals. The achieved experimental results on ear database are promising and demonstrate that the proposed method achieves higher recognition performance.
Key Words: Ear recognition; Force Transformation; Divergence; SIFT
參考文獻(References)
[1] IANNARELLI A. Ear identification, forensic identification series [M]. California: Paramount Publishing Company, 1989.
[2] 田 瑩，苑瑋琦.人耳識別技術研究綜述[J] 計算機應用研究 2007 Vo.24, No.4
TIAN Ying, YUAN Wei-Qi Survey of Human Ear Recognition [J] Application Research of Computers 2007 Vo.24, No.4
[3] 田 瑩，苑瑋琦. 基于力場轉換的人耳特征提取與識別[J] 儀器儀表學報. 2009 Feb .Vol.30 No.2
TIAN Ying, YUAN Wei-Qi Ear feature extraction and recognition based on force field transformation [J] Chinese Journal of Scientific Instrument 2009 Feb .Vol.30 No.2
[4] HURLEY D J, MARX S. Nixon, CARTER J N. Force field energy functions for image feature extraction [J]. Image and Vision Computing, 2002, 20(56):311- 317.
[5] 趙晶.基于力場轉換理論的人耳識別算法[D].遼寧沈陽:沈陽工業大學 2009
[6] HURLEY D J, MARK S. Nixon, CARTER J N. Force field feature extraction for ear biometrics [J]. Computer Vision and Image Understanding, 2005, 98(3): 491- 512.
[7] HURLEY D J, MARK S. Nixon, CARTER J N. A new Force field transform for ear and face recognition [C] , International Conference on Image Procession, 2000,25-28.
[8] 董冀媛，穆志純, 王瑜. 基于力場收斂特征的多姿態人耳識別[J], 計算機應用研究，2009 Vol.26.No.6
DONG Ji-Yuan, MU Zhi-Chun, WANG Yu .Multi-pose ear recognition based on force field convergence feature Application Research of Computers [J] 2009 Vol26 No.6
[9] 高淑欣，穆志純.人耳識別中的圖像歸一化研究[J] 控制工程2008 Vol.15, No.1
GAO Shu-Xin, MU Zhi-Chun On Image Normalization in Ear Recognition [J] Control Engineering of China 2008 Vol.15, No.1
作者簡介：崔言偉，男，2010級北方工業大學信息工程學院研究生。主要研究方向：圖像處理
