首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于改进型高斯混合模型的嵌入式实时运动检测系统
引用本文:陈龙虎,尚岩峰,梅林,刘云淮,汤志伟.基于改进型高斯混合模型的嵌入式实时运动检测系统[J].应用声学,2014,22(12).
作者姓名:陈龙虎  尚岩峰  梅林  刘云淮  汤志伟
作者单位:公安部第三研究所物联网中心,公安部第三研究所物联网中心,公安部第三研究所物联网中心,公安部第三研究所物联网中心,公安部第三研究所物联网中心
基金项目:“核高基”国家科技重大专项资助项目(No.2013ZX010033002-003);国家十二五科技支撑计划项目资助(No.2012BAH07B01)
摘    要:针对在嵌入式CPU上难以做到复杂背景下运动目标的实时检测,提出了一种基于改进型高斯混合模型的实时运动检测方案,采用改进型高斯混合模型,对高斯混合模型进行简化和结构调整,同时进行了C语言层面和CPU层级的优化,使其更合适于嵌入式平台,并详细分析了DM6446平台的软硬件设计,介绍了该算法在DM6446平台上的实现过程。实验结果表明:该系统能够有效克服外界环境变化带来的干扰,能够实时检测,可以实现多目标跟踪。

关 键 词:高斯混合模型    DSP优化    实时运动检测
收稿时间:2014/4/17 0:00:00
修稿时间:2014/5/19 0:00:00

The Embedded Real-Time Detection system of Moving Object based on Improved Gaussian Mixed Model
Shang Yanfeng,Mei Lin,Liu Yunhuai and Tang Zhiwei.The Embedded Real-Time Detection system of Moving Object based on Improved Gaussian Mixed Model[J].Applied Acoustics,2014,22(12).
Authors:Shang Yanfeng  Mei Lin  Liu Yunhuai and Tang Zhiwei
Institution:The Third Research Institute of Ministry of Pubic Security,The Third Research Institute of Ministry of Pubic Security,The Third Research Institute of Ministry of Pubic Security,The Third Research Institute of Ministry of Pubic Security,The Third Research Institute of Ministry of Pubic Security
Abstract:It is difficult to achieve real-time detection of moving object on embedded CPU, especially under complex background, so it proposes a real-time embedded detection scheme based on Gauss Mixed Model, which uses the improved Gauss mixed model, and simplifies the model. The optimized algorithm on C code level and CPU level has made the model more suitable for embedded CPU, also introduces the realization of the algorithm on the DM6446 platform and the system design of software and hardware. the experimental results show that the system can effectively overcome the interference of external environment change and realize real-time detection and multiple target tracking.
Keywords:Gauss Mixed Model  DSP optimization  real-time motion detection
点击此处可从《应用声学》浏览原始摘要信息
点击此处可从《应用声学》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号