基于嵌入式的运动目标检测与跟踪的研究与实现  

Based on Embedded in the Research and Implementation of Moving Object Detection and Tracking

作　　者：陆愿基 张龙[1] Lu Yuanji;Zhang Long(The College of Computer and Information Engineering of Tianjin Agriculture University,Tianjin, 300384,China)

机构地区：[1]天津农学院计算机与信息工程学院,天津300384

出　　处：《科技创新导报》2017年第14期209-214,216,共7页Science and Technology Innovation Herald

基　　金：天津农学院大学生创新创业计划项目(项目编号201610061182)

摘　　要：我国智能设备正处于发展的初级阶段,嵌入式平台在传统的硬件设施上的应用还有很大的发展空间。由于嵌入式系统拥有可靠、便携等特点,因此可以应用于家庭监控、智能交通等各个方面,而且近年随着ARM芯片性能的不断提高和计算机视觉算法的突破,使得图像分析技术应用到嵌入式平台。该次主要针对应用于传统视频的高速公路的车辆目标跟踪系统的研究与开发。主要研究内容及工作有:(1)高速公路的车辆目标检测跟踪系统的总体设计及软硬件平台的搭建。根据对嵌入式平台的性能和成本考虑,最后选择tiny210作为开发板,摄像头采用罗技C210,通过裁剪和移植Android系统使其适用于开发板。系统的软件框架主要分为驱动层、驱动服务层、图像分析层、应用层。利用NDK包结合C/C++和Java进行混合开发,以C/C++高性能贴近硬件的特点去写驱动层、驱动服务层、图像分析层。应用层主要使用Android的UI框架。(2)高速公路的车辆目标检测跟踪的理论基础与相关算法的实现。针对硬件平台的特性和应用在高速公路的情景下,我们采用存储空间小、计算量小、实时性好的vibe检测算法并详细介绍运动目标跟踪框架,在最后,对vibe检测算法实现并分析结果。Smart devices in our country is in the primary stage of development,the embedded platform application in the traditional hardware also has the very big development space.Due to the characteristics of embedded system with reliable,portable,so can be applied to many aspects,such as family monitoring,intelligent transportation,and in recent years,with the constant improvement of the performance of ARM chips and computer vision algorithm,makes the image analysis technology is applied to embedded platform.This mainly aimed at the application of traditional video of highway vehicle tracking systems research and development.The main research contents and work are:(1)Highway vehicle target detection tracking system’s overall design and construction of the hardware and software platform.Based on embedded platform of performance and cost considerations,the final choice tiny210as development board,the camera USES the logitech C210,by cutting and transplant the Android system is suitable for the development board.Software of the system framework consists of:driver layer,service layer,the image analysis layer,application layer.Using the NDK package of mixed with C/C++and Java development,with high performance close to the hardware characteristics of C/C++to write driver layer,service layer and image analysis.The application layer mainly using Android UI framework.(2)the highway vehicle target detection tracking theoretical basis and related implementation of the algorithm.According to the characteristics of hardware platform and application in highway scenario,we use storage space is small,small amount of calculation,good vibe real-time detection algorithm and detailed movement target tracking framework.(3)highway vehicle target detection analysis of the results of tracking system.After the results of vehicle target detection and tracking of the vehicle trajectory error analysis.

关 键 词：嵌入式 目标跟踪与检测 vibe算法 高速公路 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]