Research on Finger Vein Recognition Based on Feature Fusion of Multimode Local Encoding

Abstract

Finger vein recognition is one of the important ways of biometric recognition. Local feature coding technology is mostly used in vein recognition research. Its single coding method can easily cause loss of the main features of finger vein images. This paper proposes a multi-mode local coding operator fusion method based on the nature of local binary mode operator coding. In this manuscript, by redesigning the cyclic gradient operator [1], the cyclic gradient operator with the same dimension as the traditional LBP operator is obtained. Based on ensuring the same feature dimension, two different encodings of LBP and ICGO are obtained. Feature fusion; using FV-USM’s finger vein database, the proposed fusion feature method is compared with the traditional LBP method. The feature fusion method proposed in this paper increases the number of extracted features, although it is not time-consuming in finger vein image processing. Obvious advantages, but in terms of finger vein similarity and finger vein recognition rate, the local feature method of multi-mode encoding fusion proposed in this research is better than the traditional local feature method.