基于Gabor变换和灰度梯度共生矩阵的人耳识别研究  

Ear Recognition Based on Gabor and GGCM

作　　者：梁晓霞[1] 封筠[1] 

机构地区：[1]石家庄铁道大学计算机与信息工程学院,河北石家庄050043

出　　处：《石家庄铁道大学学报（自然科学版）》2011年第1期78-83,90,共7页Journal of Shijiazhuang Tiedao University(Natural Science Edition)

基　　金：河北省科学技术研究与发展计划项目(072135201)

摘　　要：人耳识别作为一种新兴的生物特征识别技术,具有其自身独特的优势。提出一种基于Gabor变换和灰度梯度共生矩阵的人耳身份识别方法。首先,利用Gabor变换和灰度-梯度共生矩阵融合提取人耳图像的纹理特征,然后采用K-NN分类器对特征进行分类。该方法用USTB人耳图像库做测试。实验结果表明介绍的提取人耳图像的纹理融合特征的方法优于只采用Ga-bor变换提取特征或是只提取灰度梯度共生矩阵的二次统计特征的性能。在明氏距离测度及K=1时,交叉验证识别率达到81.77%。As an emerging biometric identification technology,ear recognition has gradually received wide attention in academic research with its own unique advantages.In this paper,a novel texture feature extraction approach based on combining Gabor transform with Gray-Gradient Co-occurrence Matrix（GGCM） is presented and K-NN classification is adopted.The proposed approach is tested on USTB ear image set.The experimental results show that the ear recognition scheme fusing Gabor and GGCM are better than those based on only using Gabor transform or secondary statistical features extracted from GGCM.The best combination occurs under the Min distance measure and K=1,and the 81.77% cross-validation recognition rate is obtained.

关 键 词：人耳识别 GABOR变换 灰度-梯度共生矩阵 K-NN 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]