利用MAS实现车辆动态识别与智能跟踪

重庆交通大学学报（自然科学版） ›› 2010, Vol. 29 ›› Issue (5): 767-771.

利用MAS实现车辆动态识别与智能跟踪

曲巨宝

武夷学院数学与计算机系, 福建武夷山 354300

收稿日期:2010-05-10 修回日期:2010-06-29 出版日期:2010-10-15 发布日期:2015-01-22
作者简介:曲巨宝(1963—),男,吉林乾安人,副教授,硕士,主要从事计算机视觉、图形图像、人工智能方面研究。E-mail:qjbok@qq.com。
基金资助:
福建省自然科学基金项目( 2006 J0414) ; 福建省教育厅基金项目( JA09240) ; 武夷学院智能计算科研团队( 2009 )

Dynamic Recognition and Intelligent Track of Vehicles Using MAS

QU Ju-bao

Department of Mathematics & Computer, Wuyi Institute, Wuyishan 354300, Fujian, China

Received:2010-05-10 Revised:2010-06-29 Online:2010-10-15 Published:2015-01-22

摘要/Abstract

摘要： 针对视频图像车辆智能跟踪问题,提出了利用帧间差异积累动态矩阵进行自适应背景建模算法,采用背景差提取运动目标区域,设计了一种基于知识的多Agent智能系统进行目标分割、轮廓提取和空域滤波,增强了抗背景干扰能力,使获得的目标区域具有更好的空域连通特性;通过自适应核窗宽改进了MeanShift算法的收敛速度,利用SSD算法实现了快速初始定位。实验结果表明,该方法自治能力强,跟踪目标快速准确,实时有效。

关键词: 跟踪, MAS, 图像, 智能, 分割, MeanShift, SSD

Abstract: Aiming at the question of vehicle intelligent track ing through video image, using the frame difference to accumulate the dynamic matrix to carry on the auto-adapted background modeling algorithm is proposed. Using the background subtraction to get the motion object region, the multi-intelligent system based on the knowledge to carry out the target segmentation, the contour extraction and the spatial filtering is designed. The multi-intelligent system enhances the anti-backg round disturbance and makes the object region to get better connectivity; meanwhile, the MeanShift algorithm converg ence rate is improved through the auto-adapted nuclear window width; furthermore, the fast initial localization is realized by using the SSD(sum of squared differences) algorithm. The massive experimental results indicate that the proposed method has a strong autonomous ability and tracks object quickly, accurately and effectively.

Key words: track, MAS, image, intelligence, segmenta tion, MeanShift, SSD

中图分类号:

TP301.6

曲巨宝. 利用MAS实现车辆动态识别与智能跟踪[J]. 重庆交通大学学报（自然科学版）, 2010, 29(5): 767-771.

QU Ju-bao. Dynamic Recognition and Intelligent Track of Vehicles Using MAS[J]. Journal of Chongqing Jiaotong University(Natural Science), 2010, 29(5): 767-771.

参考文献

[1] Colombari A,Fusiello A,Murina V.Segmentation and tracking ofmultiple video objects[J].Pattern Recognition,2007,40(4):1307-1317.
[2] 王凤朝,黄树采,韩朝超.基于改进的二维Otsu法的图像分割法[J].航空计算技术,2008,38(4):4-7.
[3] 张雷,刘冀伟,王志良.固定场景下的运动检测与运动跟踪[J].微计算机信息,2006,22(25):287-289.
[4] Doucet A,Gordon N,Krishnamurthy V.Particle filter for state es-timation of jump Markov linear systems[J].IEEE Transactions onSignal Processing,2001,49(3):613-624.
[5] 张玲,蒋大永,何伟,等.基于Mean-shift的改进目标跟踪算法[J].计算机应用,2008,(28)12:3120-3122.
[6] Comaniciu D,Ramesg V,Meer P.Kernel-based object tracking[J].IEEE Transactions on Pattern Analysis and Machine Intelli-gence,2003,25(5):564-577.
[7] Colins R T.MeanShift Blob Tracking through Scale Space[C]//IEEE Computer Society Conference on Computer Vision and Pat-tern Recognition.Vancouver,Canada:IEEE Press,2003:234-240.
[8] 彭宁嵩,杨杰,刘志,等.MeanShift跟踪算法中核函数窗宽的自动选取[J].软件学报,2005,16(9):1542-1550.
[9] 陈小波,程显毅.一种基于MAS的自适应图像分割方法[J].智能系统学报,2007,2(4):80-85.
[10] 章普生,刘建平,刘国良.基于权重的形态学腐蚀改进算法[J].计算机仿真,2008,25(6):331-334.
[11] Fukunaga K,Hostetler L D.The estimation of the gradient of adensity function[J].IEEE Transactions on Information Theory,1975,2(1):32-40.
[12] Stauffer C,Grimson WE L.Adaptive Background Mixture Modelsfor Real-time Tracking[C]//Proceedings of IEEE CS Conferenceon Computer Vision and Pattern Recognition.Fort Collins,CO,USA:IEEE,1999:246-252.
[13] Collins R T.MeanShift blob tracking through scale space[J].IEEE Computer Vision and Pattern Recognition:CVPR2003,2003,1:234-240.
[14] Olfati S R.Flocking for multi-agent dynamic systems:algorithmsand theory[J].IEEE Transactions on Automatic Control,2006,51(3):401-420.
[15] 覃剑,曾孝平,李勇明.基于边界力的MeanShift核窗宽自适应算法[J].软件学报,2009,20(7):1726-1734.
[16] Jiang B A,Lu HZh.Particle filter for target tracking[J].RadarScience and Technology,2003,1(3):170-178. [17] Comaniciu D.Kernel-based object tracking[J].IEEE Transac-tions on Pattern Analysis and Machine Intelligence,2003,25(2):564-577.
[18] Raymer ML,Punch W F,Goodman E D.Dimensionality reduc-tion using genetic algorithms[J].IEEE Transactions on Evolu-tionary Computation,2000,4(2):164-171.
[19] Scharstein D,Szeliski R.A taxonomy and evaluation of dense twoframes stereo correspondence algorithms[J].Int J Comput Vi-sion,2002,47(3):37-42.