摘要
在手掌静脉识别系统中,由于人体手掌生理结构特性,容易出现图像模糊现象,导致识别性能下降。针对这一实际问题,提出一种基于子区域局部相位量化(SLPQ)的鲁棒识别方法。首先建立了LPQ特征提取的基本模型,阐述了算法流程。然后将掌脉图像分成若干子区域,提取各分区LPQ特征,并加以融合,最后利用卡方距离匹配识别。在香港理工大学、中科院自动化所和自建的掌脉图库上进行了实验测试,结果表明,当图像子区域大小为16×16像素时,该方法获得了最低的等误率,分别为0.499 2%、11.040 1%和1.356 9%,相比其他典型方法具有优势,有效提升了手掌静脉识别系统的性能,增强了系统的抗模糊性,具有实际应用价值。
In palm vein identification system,the image blur phenomenon may occur because of the physiological structure features of human palm,and this will result in poor recognization performance.Aiming at this practical problem,a robust recognization method based on sub-region local phase quantization （SLPQ） is proposed.Firstly,LPQ feature extraction basic model is established,and the process of the algorithm is described afterwards.Secondly,the palm vein image is divided into several sub-regions,and the feature of each region is extracted,then these features are fused.Finally,the chi-square distance is used for matching recognization.The presented approach was tested on the palm vein databases of Hong Kong Polytechnic University （PolyU）,Chinese Academy of Sciences Institute of Automation （CASIA） and the self-buih palm vein database.Experiment results show that the proposed method can achieve the lowest equal error rates when the image sub-region size is 16 × 16 pixels for these three image databases,which are 0.4992 ％,11.0401％ and 1.3569％,respectively.Compared with other typical algorithms,the method proposed in this paper can effectively improve the recognization and anti-blur performance of the palm vein identification system,and has practical application value.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2014年第3期543-549,共7页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(60972123)
辽宁省教育厅科学研究一般项目(L2013444)
辽宁工程技术大学博士启动基金(13-1121)资助项目
关键词
掌脉识别
特征提取
子区域
局部相位量化
palm vein recognization
feature extraction
sub-region
local phase quantization （LPQ）
