摘要
随着5G通信和物联网大数据技术的高速发展,传统的云计算模式已经越来越跟不上数据的增长速度了,边缘计算作为一种新的计算模式,表现出了很强的处理大数据和高速计算的能力.本文在“智能交通仿真系统”课题中,提出了一种适合于视频图像处理的边缘计算框架,并对传统的动目标跟踪算法进行了两点改进:(1)采用树莓派作为视频前端处理器,具有体积小,成本低,计算能力强的特点,适合边缘计算;(2)针对跟踪阶段需要采集大量样本,计算量大的缺点,采用了一种较小步长作为滑动窗的分步式图像采样方法对原压缩跟踪算法进行改进,从而减少了计算量.计算机仿真实验的结果证实了该算法在基本不影响跟踪精度的情况下提高了运算速度.
With the rapid development of 5G communications technology, IoT and big data technology, traditional cloud computing models have become increasingly unable to keep up with the growth rate of data, as a new computing model, edge computing has demonstrated a strong ability to handle big data and high-speed computing. This paper propose an edge computing framework suitable for video image processing, and two improvements to the traditional moving target tracking algorithm:(1) Raspberry Pi is used as the video front-end processor, which has the characteristics of small size, low cost, and strong computing power;(2) a step-by-step image sampling method with a smaller step size as a sliding window is used to improve the original compression tracking algorithm, thereby reducing the amount of calculation. The results of computer simulation experiments show that the algorithm improves the operation speed without affecting the tracking accuracy.
作者
刘浩
卿粼波
宗江琴
陈虹君
LIU Hao;QING Lin-Bo;ZONG Jiang-Qin;CHEN Hong-Jun(College of Electronic Information Engineering,Chengdu Jincheng College,Chengdu 611731,China;College of Electronic Information Engineering,Sichuan University,Chengdu 610065,China;Jiangxi Information Technology School,Nanchang 330006,China;Sichuan Expert Workstation of Chengdu Jincheng University,Chengdu 611731,China)
出处
《四川大学学报（自然科学版）》
CAS
CSCD
北大核心
2022年第6期93-99,共7页
Journal of Sichuan University(Natural Science Edition)
基金
四川省科技厅重点研发项目(22ZDYF0724)。
关键词
边缘计算
树莓派
稀疏表示
压缩跟踪
分步式采样
Edge computing
Raspberry
Sparse representation
Compression tracking
Stepwise sampling
