Finger vein recognition system based on shallow convolutional neural network

Abstract

The invention provides a finger vein recognition system based on a shallow convolutional neural network. The finger vein recognition system comprises three convolution modules, and two full connectionlayers, each convolution module comprises a convolution layer, a pooling layer and a batch processing standardization layer, so that the system is a shallow convolution neural network framework, andthe problem of overfitting caused by excessive feature extraction of the deep convolution neural network is avoided; the network has enough capacity of extracting finger vein features and has strongerrecognition performance; the system is short in training period and easy to converge, so that the practicability of a closed-set architecture can be realized; the mapping relationship between the first full connection layer and the second full connection layer in the open set architecture can be adjusted by modifying a loss function, and on the basis, open set identification is realized; switching between a closed-set architecture and an open-set architecture can be achieved only through different loss functions, and finger vein recognition of the closed-set architecture and finger vein recognition of the open-set architecture cannot be achieved at the same time through the shallow convolutional neural network framework.

Description

technical field [0001] The present application relates to the technical field of biometrics, and in particular to a finger vein recognition system based on a shallow convolutional neural network. Background technique [0002] Finger vein recognition is to use the vein distribution image in the finger to identify the identity, specifically based on the fact that the blood flowing in the finger can absorb light of a specific wavelength, and use a specific wavelength optical fiber to irradiate the finger to obtain a clear image of the finger vein, and then analyze the image Analyze and process to obtain the biological characteristics of the finger vein, and compare the obtained biological characteristic information with the pre-registered finger vein characteristics to complete the identification; with the emergence of powerful deep learning algorithms, more and more The researchers introduced deep learning into finger vein recognition; finger vein recognition based on deep lea...

AI Technical Summary

Problems solved by technology

[0005] This application provides a finger vein recognition system based on a shallow convolutional neural network to solve the technical problem that the deep convolutional network is overfitting and cannot simultaneously realize a closed-set structure and an open-set structure

Images