基于大数据的边境人员图像修复应用研究
Research on Image Restoration of Border Personnel Based on Big Data
王文婕 1秦振凯 2罗起宁 1陈广成3
作者信息
- 1. 广西警察学院信息技术学院,广西南宁 530028
- 2. 广西警察学院信息技术学院,广西南宁 530028;广西警察学院 公安大数据现代产业学院,广西南宁 530028
- 3. 苏州科达科技股份有限公司,江苏苏州 215501
- 折叠
摘要
随着人脸识别和视频监控技术的快速进步,人像检测在边境安全中变得至关重要.图像质量,常受模糊和噪声影响,对监控效率构成挑战.为此,文章引入一种基于Transformer的模糊人脸修复网络,旨在改善受损图像质量,提升边境监控系统的性能.通过在Columbia Gaze数据集上进行实验,评估了模型在关键性能指标LPIPS和SSIM上的表现.结果显示,该网络显著提高了图像清晰度,为边境人员图像的实时预警提供了坚实技术支持,显著提升了监控系统的可靠性和工作效率,展现了在边境安全监控领域的应用潜力.
Abstract
With the rapid progress of face recognition and video surveillance technology,portrait detection has become crucial in border security.Image quality,often affected by blur and noise,poses a challenge to the monitoring efficiency.To this end,this study introduces a Transformer-based fuzzy face repair network,aiming to improve the damaged image quality and enhance the performance of the border monitoring system.The performance of the model on the key per-formance indicators LPIPS and SSIM was evaluated by performing experiments on the Colum-bia Gaze dataset.The results show that the network significantly improves the image clarity,provides solid technical support for the real-time warning of the images of border personnel,sig-nificantly improves the reliability and working efficiency of the monitoring system,and shows the application potential in the field of border security monitoring.
关键词
大数据/边境人员/Transformer人脸图像修复/鲁棒性Key words
deep learning/image restoration/border security引用本文复制引用
出版年
2024
NETL
NSTL
万方数据