基于嵌入式的智能远程视频监控系统设计  

Design of an embedded intelligent remote video monitoring system

作　　者：孙留存 苏卫江 库斯达 高永强 王政 SUN Liucun;SU Weijiang;KU Sida;GAO Yongqiang;WANG Zheng(China Green Development Investment Group,BeiJing 100020,China)

机构地区：[1]中国绿发投资集团有限公司,北京100020

出　　处：《自动化与仪器仪表》2024年第8期163-167,共5页Automation & Instrumentation

摘　　要：为了实现更高效、准确的远程监控,研究提出一种基于嵌入式的智能远程视频监控系统设计方法,并对人脸识别和车牌识别算法进行改进。该系统的硬件部分包括嵌入式开发板、摄像头、无线网卡、存储卡以及开发板集成的网口和接口等设备。测试结果显示,视频的传输延迟均在0~100 ms范围内,视频传输稳定性均在90%~100%范围内。说明系统能够稳定地传输视频数据,保证了监控的准确性和可靠性。研究提出的基于深度可分离中心差分卷积网络算法在人脸和车牌识别中的准确率分别为98.9%和99.2%,均优于卷积神经和中心差分卷积网络算法。智能远程视频监控系统的系统功能评分为92分,相比于原系统提升了33分。结果表明,该系统具有良好的实时监控性能和视频分析能力,可以提高监控的效率和准确性,为监控应用提供了一个可靠的解决方案。In order to achieve more efficient and accurate remote monitoring,a design method for an embedded intelligent remote video monitoring system is proposed,and facial recognition and license plate recognition algorithms are improved.The hardware of the system includes embedded development boards,cameras,wireless network cards,storage cards,as well as integrated network ports and interfaces on the development board.The test results show that the video transmission delay is within the range of 0-100 ms,and the video transmission stability is within the range of 90%-100%.The system can stably transmit video data,ensuring the accuracy and reliability of monitoring.The accuracy of the proposed deep separable center difference convolutional network algorithm in facial and license plate recognition is 98.9%and 99.2%,respectively,which are superior to convolutional neural and center difference convolutional network algorithms.The system function score of the intelligent remote video monitoring system is 92 points,which is 33 points higher than the original system.The results show that the system has good real-time monitoring performance and video analysis ability,which can improve the efficiency and accuracy of monitoring and provide a reliable solution for monitoring applications.

关 键 词：嵌入式 树莓派4B 远程 视频监控 高清摄像头 视频分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]