欢迎访问《图学学报》 分享到：
图学学报

图学学报

• 图像处理与计算机视觉 • 上一篇    下一篇

融合几何信息和方向信息的三维掌纹识别方法

  

  1. (1. 东南大学自动化学院，江苏 南京 210096；
    2. 东南大学复杂工程系统测量与控制教育部重点实验室，江苏 南京 210096)
  • 出版日期:2020-06-30 发布日期:2020-08-18
  • 基金资助:
    国家自然科学基金项目(51475092，61462072)；江苏省自然科学基金项目(BK20181269)；深圳市知识创新计划基础研究项目(JCYJ2018030
    6174455080)

Fusion of geometric and orientation information for 3D palmprint recognition

  1. (1. School of Automation, Southeast University, Nanjing Jiangsu 210096, China;
    2. Key Laboratory of Measurement and Control of Complex Engineering System, Ministry of Education, Southeast University, Nanjing Jiangsu 210096, China)
  • Online:2020-06-30 Published:2020-08-18

PDF (PC)

100

摘要： 针对三维掌纹特征表示的鲁棒性和准确性问题，提出一种融合曲面的几何特征和
方向特征的三维掌纹识别方法。基于现有的曲面类型编码提取掌纹几何特征的基础上，提出使
用基于形状指数的编码来共同表达三维掌纹的几何特征，从而有效减少由阈值所引起的错误编
码带来的准确性上的影响。此外，提出一种多尺度的改进竞争编码来表达掌纹的方向特征。在
决策层，使用基于多字典的协同表示框架融合上述几何特征和方向特征以完成掌纹识别。在公
开的三维掌纹数据集上的大量实验表明，所提方法可以在保持较低计算复杂度的同时实现最佳
的识别精度。

关键词: 三维掌纹识别, 生物特征, 形状指数编码, 改进竞争编码, 协同表示

Abstract: In order to improve the robustness and accuracy of the feature representation of 3D
palmprint, a method integrating the geometric and directional features of curved surfaces was
proposed. Based on the existing method using the surface type (ST)-based coding to extract geometric
features of a 3D palm, we proposed to use the shape index (SI)-based coding to jointly characterize
the geometric features of 3D palmprints. This operation can effectively reduce the impact on accuracy
brought by the error encoding caused by the threshold. Moreover, we proposed a multi-scale modified
competitive coding (MSMCC) to characterize the orientation features. The multi-dictionary
collaborative-representation (CR)-based framework was employed to merge the geometric and
orientation features into the decision level to perform identification. Extensive experiments on the
public 3D palmprint database prove that the proposed method can achieve an optimal rank-1
recognition accuracy while maintaining a relatively low computational complexity.

Key words: 3D palmprint recognition, biometrics, shape index coding, modified competitive coding;
collaborative representation