基于ZYNQ的实时运动目标检测系统设计  

Design of a Real-Time Moving Target Detection System Based on ZYNQ

作　　者：肖明艳 XIAO Mingyan(School of Information and Software Engineering,Kunshan Dengyun College of Science and Technology,Suzhou 215300,China)

机构地区：[1]昆山登云科技职业学院信息与软件工程学院,江苏苏州215300

出　　处：《苏州市职业大学学报》2025年第1期48-53,共6页Journal of Suzhou Vocational University

基　　金：江苏省职业院校学生创新创业培育计划项目(G-2023-0921)。

摘　　要：传统视频监控系统在实时运动目标检测时,存在目标检测不完整和目标检测错误的局限性。本研究基于体积小、实时性高的需求,提出了一种将动态三帧差分法与Sobel边缘检测算法结合的实时目标检测方法,并基于ZYNQ构建了视频采集模块、图像处理模块、图像缓存模块、视频显示模块和目标地址读取程序,可提高检测的准确性与实时性。在CDnet数据集与实拍场景上进行实验验证,结果表明该系统能够在不同场景下有效、准确地检测实时运动目标。Traditional video surveillance systems have the limitations of incomplete target detection and incorrect target detection during real-time moving target detection.Under the requirements of small volume and high real-time performance,this paper proposes a real-time target detection method that combines the dynamic three-frame difference method with the Sobel edge detection algorithm.A system has been designed based on ZYNQ,incorporating modules for video capture,image processing,image buffering,video display,and a target address reading program,which can improve the accuracy and real-time performance of detection.Experiments on CDnet datasets and real-shot scenarios show that the system can effectively and accurately detect real-time moving objects in different scenarios.

关 键 词：ZYNQ 动态三帧差分法 目标检测 实时图像处理 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置] G640[自动化与计算机技术—控制科学与工程]