您的位置:首页  > 论文页面

一种运动车辆识别方法及在Davinci平台上的实现

发表时间:2010-05-31  浏览量:1351  下载量:652
全部作者: 唐小林,曹广忠
作者单位: 深圳大学自动化研究所;深圳市传感器技术重点实验室,深圳大学物理科学与技术学院
摘 要: 针对运动车辆识别算法对实时性的要求,设计一种基于嵌入式Davinci平台的运动车辆识别方法。首先,设计出一种运动车辆识别算法。该算法通过在图像中设置虚拟检测线实现运动车辆检测,采用背景差法提取出运动车辆图像,通过对车辆图像的边界追踪求出车长及车宽,且实时地自适应更新背景,利用车长及车宽信息完成车型识别。Matlab仿真结果表明:该算法能有效识别路面运动车辆。基于Davinci的软件框架,研究并将该算法移植到Davinci平台上,给出基于Davinci平台的嵌入式车辆识别系统。实验结果表明:该系统能有效识别路面运动车辆,识别时间短、功耗低、性能稳定,具有一定应用价值。
关 键 词: 车辆识别;车辆检测;虚拟检测线;背景差;Davinci平台
Title: A method for moving vehicle recognition and it�s realization on Davinci platform
Author: TANG Xiaolin, CAO Guangzhong
Organization: Institute of Automation, Shenzhen University;Shenzhen Key Laboratory of Sensor Technology, College of Physics Science & Technology, Shenzhen University
Abstract: Aiming at the real�time request of moving vehicles recognition, this paper proposes an moving vehicle recognition method based on embedded Davinci platform. The paper designs a moving vehicle recognition algorithm, which combines virtual detection line to computer vision. It applies background subtraction to separate moving vehicle from background, gets vehicle's length and width from the contour, updates background in real time, and implements vehicle type reorganization by use of the length and width. The simulation result on Matlab shows that the algorithm can effectively recognize the moving vehicle. By implementation of this algorithm on Davinci platform, this paper provides an embedded vehicle recognition system. Experimental results show that this system is capable of effectively recognizing moving vehicles, has advantages of short recognition time, low energy consumption and stable performance, and has application value.
Key words: vehicle recognition; vehicle detection; virtual detection line; background subtraction; Davinci platform
发表期数: 2010年5月第10期
引用格式: 唐小林,曹广忠. 一种运动车辆识别方法及在Davinci平台上的实现[J]. 中国科技论文在线精品论文,2010,3(10):1025-1030.
 
0 评论数 0
暂无评论
友情链接