基于可控方向直方图算法的掌纹图像特征提取技术  

Palmprint image feature extraction technology based on controllable direction histogram algorithm

作　　者：陈小祥 CHEN Xiao-xiang(College of Information Technology,Anqing Vocational&Technical College,Anqing 246003,Anhui,China)

机构地区：[1]安庆职业技术学院信息技术学院,安徽安庆246003

出　　处：《贵阳学院学报（自然科学版）》2022年第1期10-13,34,共5页Journal of Guiyang University：Natural Sciences

基　　金：2018年安徽省高校自然科学研究项目“教学过程支撑系统的开发与应用”(项目编号:KJ2018ZD067)。

摘　　要：研究基于掌纹图像特征提取技术对HOG梯度方向直方图算法和方向可控滤波器方法进行了阐述,并以等错误率EER作为评价指标,曼哈顿距离和卡方距离作为相似性度量标准,参考PolyU II数据库,对训练图像和测试图像在不同算法下的识别率进行比较。研究结果显示:HOS算法下的卡方距离EER值和曼哈顿距离EER值分别为0.6563和0.5308,且远小于其他算法,这说明HOS算法在掌纹图像特征提取上具有较好的工作性能,这为目前的生物特征图像识别算法提供了一种新的组合思路。based on the palmprint image feature extraction technology,the hog gradient direction histogram algorithm and the direction controllable filter method are described.Taking the equal error rate eer as the evaluation index,Manhattan distance and chi square distance as the similarity measurement standard,and referring to PolyU II database,the recognition rates of training image and test image under different algorithms are compared.The results show that the chi square distance EER and Manhattan distance eer under HOS algorithm are 0.6563 and 0.5308 respectively,which are much smaller than other algorithms,which shows that HOS algorithm has good performance in palmprint image feature extraction,which provides a new combination idea for the current biometric image recognition algorithms.

关 键 词：可控方向直方图 掌纹特征 掌纹 识别率 

分 类 号：TP212.9[自动化与计算机技术—检测技术与自动化装置]