A finger vein authentication method based on long-term recurrent convolutional neural network

Abstract

The invention provides a finger vein authentication method based on a long-term recursive convolutional neural network. Firstly, the finger vein image is marked and a label sequence is created, and then the long-term recursive convolutional neural network is trained to perform texture identification and spatial dependence on the finger vein separation. Representation; train the probabilistic support vector machine to calculate the probability that the corresponding pixel belongs to the vein; finally, calculate the non-overlapping area of ​​any two images of the same sample at different spatial displacements through the enhanced Hamming distance, so that the All augmented images are pairwise matched. The method can quickly and efficiently record and identify finger veins, and can effectively improve the accuracy of finger vein authentication.

Description

technical field [0001] The invention belongs to the technical field of finger vein authentication, in particular to a finger vein authentication method based on a long-term recursive convolutional neural network. Background technique [0002] Biometric identification technology is a technology that uses human biological characteristics or behavioral characteristics for identity authentication. The behavioral characteristics mainly include signature, voice, gait, etc. Biological characteristics can be divided into external biological characteristics (fingerprint, palm type, iris vision, etc.) Face shape, etc.) and internal biological characteristics (finger veins, dorsal veins and palm veins, etc.) Among them, fingerprint recognition is widely used due to its uniqueness, stability, and ease of use. However, in fingerprint identification, users must be required to keep their fingers clean and smooth when entering fingerprints, because any dirt or stains on the fingerprints can...

AI Technical Summary

Problems solved by technology

For this reason, in the process of finger vein recognition, since the collection of finger vein images is affected by many factors, such as ambient light, ambient temperature, and light scattering, finger vein recognition still faces severe challenges.

These factors are difficult to control and overcome in practical applications, so the collected images contain many blurred areas in which the distinction between the finger vein features and the background is poor. Regions can lead to a significant reduction in authentication accuracy

At present, the method based on manual features is difficult to effectively extract finger vein pattern information, resulting in limited authentication efficiency and accuracy.

Images