指静脉识别技术研究

Research on Finger Vein Recognition Technology

目的分析指静脉识别技术的发展现状,有助于相关学者快速把握该领域的发展动态。方法围绕指静脉识别技术流程展开,研究当下指静脉图像的采集方式、图像预处理方法、特征提取及匹配的相关算法,并分析各阶段面临的技术难点及可以突破的领域。从多视点融合、多实例融合、多算法融合、多模态融合角度深入分析了近年来兴起的指静脉融合技术,并指出当下指静脉融合技术面临的瓶颈。结论①图像增强领域,对比度和纹理增强研究已经很深入,细小静脉缺失研究还很初步,可以在此领域有所新突破,此外,ROI提取去除的手指静脉也包含着有效信息,目前没有得到有效的研究利用。②特征提取领域,纹理特征较几何分形特征具有较好的局部偏差容忍度,特征匹配领域,深度学习将会开启模式识别的新纪元。③指静脉相关融合技术是当下的研究热点,而其中的多模态融合技术将备受学者青睐。④数据库规模是目前限制指静脉识别技术及相关研究的主要因素。

Objective This paper analyzes the development status of finger vein recognition technology,which is helpful for relevant scholars to grasp the development trend of this field quickly.Methods This paper focuses on the technical process of the finger vein recognition,and studies the acquisition method,image preprocessing method,feature extraction and matching algorithms,and analysis the technical difficulties faced by each stage and areas where breakthroughs can be made.In addition,this paper also analyzes the emerging finger vein fusion technology in recent years from multi-point fusion,multi-instance fusion,multi-algorithm fusion,and multi-modal fusion and points out the bottleneck in this research field.Conclusion①In the field of image enhancement,the research on contrast and texture enhancement has been studied in-depth,and the study on the loss of tiny veins is still very preliminary,which can make a new breakthrough in this field.In addition,the finger vein areas removed during ROI extraction also contain valid information,which has not been effectively studied and utilized at present.②In the filed of feature extraction,texture features have better tolerance of local deviation than geometric features.In the filed of feature matching,depth study will open a new era of pattern recognition.③Finger vein fusion technology is a hot research topic at present,among which multimodal fusion will be favored by many scholars.④At present,database size is the main factor limiting finger vein identification technology and related research.