首页|基于人脸识别的多轨道融合车站电子车票自动化验证研究

基于人脸识别的多轨道融合车站电子车票自动化验证研究

扫码查看
在嘈杂的车站环境中,验证设备可能受到周围噪音与光线的干扰,导致电子车票自动化验的精度较低.针对上述情况,研究一种基于人脸识别的多轨道融合车站电子车票自动化验证方法.针对原始人脸图像特征信息不明显或丢失的情况,实施人脸光照校正和人脸遮挡恢复处理,提高人脸图像质量.提取人脸图像LBP特征并统计其直方图,得到LBP特征直方图特征向量.基于卷积神经网络构建识别模型,从车票信息库中识别出与之对应的人脸图像,由此完成电子车票自动化验证.结果表明:所研究方法的应用下,F1值相比更大,由此说明该方法的验证能力更强,更能保证在多轨道融合车站电子车票自动化验证中的精度.
Research on Automatic Verification of Electronic Tickets for Multi track Fusion Stations Based on Face Recognition
In a noisy station environment,the validation equipment may be disturbed by ambient noise and light,resulting in low accuracy of electronic ticket automatic test.In response to the above situation,a multi track fusion station electronic ticket automated verification method based on facial recognition is studied.In response to the situation where the feature information of the original fa-cial image is not obvious or lost,the implementation of facial illumination correction and facial occlusion restoration processing can improve the quality of the facial image.Extract LBP features from facial images and calculate their histograms to obtain LBP feature histogram feature vectors.The recognition model is constructed based on Convolutional neural network,and the corresponding face image is recognized from the ticket information database,thus completing the automatic verification of electronic tickets.The results indicate that under the application of the studied method,the F1 value is larger,indicating that the validation ability of this method is stronger and can better ensure the accuracy of electronic ticket automation validation in multi track fusion stations.

facial recognitionmulti track fusion stationpre processingLBP featuresconvolutional neural networkelectron-ic ticket verification

黄庆贵、李海培、王峰

展开 >

中铁第一勘察设计院集团有限公司,西安 710043

人脸识别 多轨道融合车站 预处理 LBP特征 卷积神经网络 电子车票验证

中国智慧工程研究会"十四五"规划重点项目

JYK9025

2024

自动化与仪器仪表
重庆工业自动化仪表研究所,重庆市自动化与仪器仪表学会

自动化与仪器仪表

CSTPCD
影响因子:0.327
ISSN:1001-9227
年,卷(期):2024.(8)