基于边缘计算的鱼眼相机高空抛物检测系统  

High-Altitude Parabolic Object Detection System of Fisheye Camera Based on Edge Computing

作　　者：林诗妍 刘宇翔 魏宏安[1] LIN Shiyan;LIU Yuxiang;WEI Hong'an(College of Physics and Information Engineering,Fuzhou University,Fuzhou 350108,China)

机构地区：[1]福州大学物理与信息工程学院,福建福州350108

出　　处：《福建师范大学学报（自然科学版）》2023年第6期52-62,共11页Journal of Fujian Normal University：Natural Science Edition

基　　金：国家自然科学基金资助项目(61671152)。

摘　　要：高空抛物引发的人身财产损失案件频发,并且抛物行为迅速而突然,目击者少且不稳定,这导致了案件的取证工作变得极为复杂.针对此类现象,提出一种基于边缘计算的鱼眼相机高空抛物检测系统,主要分为图像视频采集、终端动目标检测以及边缘端目标识别3个模块,在终端将高空抛物视频矫正畸变后,利用三帧差法对抛物者与抛物对象进行标记,随后发送视频序列至边缘端进行目标识别.实验结果的平均精确率均值为0.994,结果表明该系统能够有效检测高空抛物行为.Incidents of personal and property damages caused by objects thrown from high alti-tudes have been occurring frequently.The sudden and rapid nature of such incidents,combined with the scarcity and instability of witnesses,has made the process of evidence collection highly complex.To address this phenomenon,this paper proposes a high-altitude parabolic object detection system of a fisheye camera based on edge computing,which comprises three modules:image and video capture,on-device moving object detection,and edge-side target recognition.After correcting the distortion in the fisheye images at the terminal,the system employs the three-frame difference method to label the thrower and the thrown object and subsequently sends the video sequence to the edge for target recognition.Experimental results demonstrate an average precision recall mean of 0.994,indicating that the system effectively detects high-altitude object throwing.

关 键 词：鱼眼相机 棋盘格标定 边缘计算 目标检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]