Finger vein recognition method based on deformation detection and correction based of convolutional neural network

Abstract

The invention discloses a finger vein recognition method based on deformation detection and correction of a convolutional neural network, and aims to solve the problem that an existing finger vein recognition algorithm is poor in recognition performance under the condition that a finger is seriously deformed. The method carefully analyzes the principle of finger vein deformation correction and introduces the convolutional neural network into a finger vein correction and recognition algorithm, and pays attention to training a recognition model meeting performance requirements as much as possible under the condition of a small data set. According to the method, deformation correction can be accurately carried out on a finger vein image and effective features are extracted to carry out identification under the condition that the finger is deformed irregularly. According to the method, the finger vein deformation correction process is fully considered, the convolutional neural network is used for finger vein deformation correction, and the efficient feature extraction network is used for extracting and recognizing finger vein features. Meanwhile, an original finger vein image is directly used as the input of the method, complex preprocessing operation is not needed, and a good recognition effect can be obtained under a small-scale data set.

Description

technical field [0001] The invention relates to a method for detecting and correcting deformation of finger vein images and identifying veins, which belongs to the field of human biological characteristics. Background technique [0002] Biometric technology is a security technology that uses the unique biological characteristics of the human body for identification. Biological characteristics refer to physiological characteristics or behaviors that are unique (distinguishable from others), universal (all normal people have), stable (no mutations), and descriptive (able to be extracted and stored), such as fingerprints , palm prints, iris, veins, face, gait, EEG, etc. Compared with traditional identity authentication technology, biometric technology is increasingly valued by various research institutions and even countries because of its safer, more convenient, and more stable characteristics, and it is widely used in various industries. [0003] Existing finger vein recogn...

AI Technical Summary

Problems solved by technology

Aiming at the problem that the existing finger vein recognition algorithm has poor recognition performance when the finger is severely deformed, this method carefully analyzes the principle of finger vein deformation correction and introduces convolutional neural network into the finger vein correction and recognition algorithm. In the case of a small data set, train a recognition model that meets the performance requirements as much as possible

Images