摘要
车联网视频采集终端在复杂环境下会出现准确率较低的问题,对此提出基于神经网络的车载视频采集终端车牌号码识别算法。依托移动边缘计算优势,在车联网侧添加边缘层MobileNet可分离卷积核,在采用滤镜算法对采集图像进行数学形态学预处理后,结合云端的深度神经网络,设计一种基于车联网边缘计算的车牌号识别算法。实验结果表明,该方法较传统方法有着更高的准确率与速率,更好满足了公安应急处突中移动车辆实时定位的要求。
To solve the problem that the accuracy rate of the video acquisition terminal in the internet of vehicles is low in the complex environment,a vehicle video acquisition terminal license plate number recognition algorithm based on neural network was proposed.According to the advantages of mobile edge computing,the edge layer MobileNet separable convolution kernel was added to the side of the internet of vehicles.After the mathematical morphology preprocessing of the collected images by using filter algorithm,a vehicle plate number recognition algorithm based on the edge computing of the internet of vehicles was designed by combining the deep neural network of the cloud.Experimental results show that the proposed method has higher accuracy and speed than traditional methods,and better meets the real-time positioning requirements of moving vehicles in emergency response.
作者
李猛坤
柯正轩
于定荣
张建林
贾军营
刘利峰
LI Meng-kun;KE Zheng-xuan;YU Ding-rong;ZHANG Jian-lin;JIA Jun-ying;LIU Li-feng(School of Management,Capital Normal University,Beijing 100089,China;Beijing Shenzhou Aerospace Software Technology,China Aerospace Science and Technology Corporation,Beijing 100094,China;Institute of Chinese Materia Medica,China Academy of Chinese Medical Sciences,Beijing 100700,China;Fengchi Research Institute,Shenyang Fengchi Software Co.LTD,Shenyang 110167,China;Beijing Yucai Technology Co.LTD,Yucai Technology Headquarters,Beijing 100040,China)
出处
《计算机工程与设计》
北大核心
2021年第11期3151-3157,共7页
Computer Engineering and Design
基金
全国高等院校计算机基础教育研究会计算机基础教育教学研究基金项目(2020-AFCEC-074)。
关键词
边缘计算
车牌号识别
深度可分离神经网络
机器学习
图像处理
edge calculation
license plate identification
depth separable neural network
machine learning
image processing
