Finger vein recognition method based on individualized weight

Abstract

The invention discloses a finger vein recognition method based on individualized weight, and the method can be used for effectively overcoming negative influence of noise levels on recognition accuracy and improving the recognition performance and robustness of a finger vein recognition system. The method comprises a training process and a recognizing process, wherein the training process comprises the following steps of: firstly preprocessing training images; then respectively extracting LBP (local binary pattern) features of the training images; and finally obtaining a weight bitmap W through training. The recognizing process comprises the following steps of: preprocessing a tested sample image firstly; then carrying out LBP feature extraction; and finally calculating the hamming distance between the tested sample image and a database template, and determining a recognition result according to a set threshold, wherein if figures on two sides of an XOR operator of which the symbol isshown in the specification are identical, the result is '0', otherwise, the result is '1', and if DA is less than the set threshold theta, the tested image belongs to class A and theta is 0.15.

Description

technical field [0001] The invention relates to the field of finger vein recognition, in particular to a finger vein recognition method based on individualized weights. Background technique [0002] Finger vein recognition is a new biometric technology with good development prospects. The key to finger vein recognition is how to accurately extract the vein network, and then perform feature extraction and matching on this basis. In order to overcome the influence of low-quality finger vein images on the recognition results, a feature extraction and matching method based on bit patterns was proposed. The basic principle is to firstly do some preprocessing on the collected finger vein images, including image enhancement, size normalization, etc., and then extract features based on bit patterns from the preprocessed images, such as LBP features, LDP features, etc. These Features are generally referred to as binary codes. Finally, the Hamming distance between the binary codes i...

AI Technical Summary

Problems solved by technology

[0003] Traditional finger vein recognition based on bit patterns often regards the contribution of each bit of binary code to the final recognition as equal when calculating the Hamming distance. In this way, the existence of some noise bits usually leads to a reduction in the recognition rate.

Images