Dorsal hand vein recognition via hierarchical combination of texture and shape clues

Di Huang, Xiangrong Zhu, Yunhong Wang, Dapeng Zhang

Research output: Journal article publicationJournal articleAcademic researchpeer-review

36 Citations (Scopus)

Abstract

Within the domain of biometrics, the pattern of dorsal hand vein has received increasing attentions during recent years due to its properties of being universal, unique, stable, and contactless, in particular the simplicity of liveness detection and difficulty of forging. This paper presents an effective and efficient approach to dorsal hand vein recognition. In contrast to the majority of existing methods, which either make use of texture features or shape features, this method highlights that their joint use improves the distinctiveness of the dorsal hand vein image so that it can be better distinguished from the ones of other subjects. A novel shape representation method is proposed to describe the geometry structure of the venous network, by integrating both local and holistic analyses. Its similarity measurement is then hierarchically combined with that from the texture modality, generated by the well reputed texture operator, namely Local Binary Patterns (LBP), for decision making. To validate the proposed approach, extensive experiments are conducted on NCUT, one of the largest publicly available dorsal hand vein databases, and the results achieved are superior to the state of the art ones so far reported in the literature, which demonstrates its effectiveness.
Original languageEnglish
Pages (from-to)815-828
Number of pages14
JournalNeurocomputing
Volume214
DOIs
Publication statusPublished - 19 Nov 2016

Keywords

  • Biometrics
  • Fusion
  • Hand vein recognition
  • Shape modeling
  • Texture representation

ASJC Scopus subject areas

  • Computer Science Applications
  • Cognitive Neuroscience
  • Artificial Intelligence

Cite this