二进制鲁棒不变尺度特征在非接触掌纹识别中的应用  

Application of binary robust invariant scalable keypoints in non-contact palmprint recognition

作　　者：林森[1,2] 苑玮琦[2] 宋辉[2] 

机构地区：[1]辽宁工程技术大学电子与信息工程学院,葫芦岛125105 [2]沈阳工业大学视觉检测技术研究所,沈阳110870

出　　处：《仪器仪表学报》2013年第12期2785-2792,共8页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(60972123);高等学校博士学科点专项科研基金(20092102110002);辽宁省博士启动基金(20131078);沈阳市科技计划(F10-213-1-00)资助项目

摘　　要：针对目前主流的非接触掌纹识别系统鲁棒性较差、识别效果不佳的问题,提出一种基于二进制鲁棒不变尺度特征(BRISK)的掌纹识别方法。首先建立了BRISK尺度空间特征提取模型和采样模式,阐述了算法步骤。然后利用BRISK算法提取非接触掌纹图像的稳定特征点,计算掌纹间的匹配点数,并以此进行识别,同时分析和给出了特征阈值的选择过程。最后在基于实际环境建立的、通用的IITD非接触掌纹库上进行测试。结果表明,该方法正确识别率可达98.124 2%,识别时间仅为0.012 1 s,进一步提升了非接触掌纹识别系统的识别效果,增强了系统的鲁棒性,方法速度快且效率高,能够满足在线识别的要求,具有应用价值。Aiming at the problems that current mainstream non-contact palmprint identification systems have unsatis- fied robustness and recognition effect, a palmprint recognition method based on binary robust invariant scalable key- points （BRISK） is proposed. First of all, the BRISK scale-space feature extraction model and sampling pattern are established, and the algorithm procedures are described afterwards. Secondly, the stable feature points of non-contact palmprint images are extracted with the BRISK algorithm,and the number of the palmprint matching points is calcu- lated, with which the palmprint recognition is conducted. Meanwhile, the selection process of the feature threshold is analyzed and given. The presented method was tested on the general IITD non-contact palmprint database established based on the actual environment. Experimental results show that the correct reeognition rate of the proposed method can reach 98. 124 2% ,and the recognition time is only 0. 012 1 s. The method illustrated in this paper can further improve the robustness and recognition effect of the non-contact palmprint identification system;the method features fast speed and high efficiency, can satisfy the requirements of online palmprint identification and has considerable ap- plication value.

关 键 词：掌纹识别 特征提取 非接触 二进制鲁棒不变尺度特征 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]