基于特征融合的非接触式掌纹识别  

Contactless Palm Print Recognition Based on Feature Fusion

作　　者：莫文博 MO Wenbo(School of Information Engineering,Quzhou College of Technology,Quzhou Zhejiang 324000,China)

机构地区：[1]衢州职业技术学院信息工程学院,浙江衢州324000

出　　处：《信息与电脑》2023年第1期179-181,共3页Information & Computer

基　　金：衢州职业技术学院2021年度校级科研项目(项目编号:QZYY2106)。

摘　　要：由于单一的特征描述掌纹中的所有信息,容易导致识别准确率低,因此提出一种边缘特征与纹理特征相结合的掌纹识别方法。先使用整体嵌套边缘检测模型和Gabor滤波器分别提取掌纹的边缘特征和纹理特征,再通过特征融合方法融合两种特征。最后,将融合后的特征图像输入到卷积神经网络,并得出分类结果。通过在不同的公开掌纹数据集上进行对比实验,结果表明本文方法取得了良好的结果。Because a single feature describes all the information in the palmprint, the recognition accuracy is low. This paper presents a palmprint recognition method combining edge feature and texture feature. The whole nested edge detection model and Gabor filter were used to extract the edge features and texture features respectively, and the two features were fused by feature fusion method. Finally, the fused feature images are input into the convolutional neural network and the classification results are obtained. The comparison experiments on different public palmprint data sets show that the proposed method achieves good results.

关 键 词：边缘特征 纹理特征 深度学习 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]