地铁站出入通道多摄像机移动目标跟踪研究
|
摘要
为及时发现、预警、防控进入地铁站的可疑人员及其行为,实现多摄像机之间的移动目标传递十分关键。为此,以行人进站乘车的必经之路——地铁站出入通道为背景,以安装其间的相邻2台摄像机的监控图像为研究对象,选用ViB e算法针对2台固定有效监测视域且无重叠区域的摄像机建立移动目标背景检测模型。当有行人出现时,建立移动目标模板,从行人的移动方向、身高宽度比、行走速度、空间颜色直方图4个方面实施特征匹配。实验结果表明:在2台摄像机之间顺利实现了对行人的移动目标传递和轨迹跟踪,跟踪准确率高于在单摄像机上直接使用SIFT算法。
In order to discover,warn,prevent and control the suspicious people and their behavior entering the subway station in time,the key is to realize the moving target transmission between multiple cameras. Therefore, the access channel of subway station is selected as this study background,and the monitoring image of two adjacent cameras installed in the access channel is taken as the research object. Vi Be algorithm is used to build the background detection model of moving target in two cameras with fixed and non-overlapping detection horizons. When pedestrians appear,the moving target template can be established to implement feature matching from the four aspects of pedestrian movement direction,height-width ratio,walking speed and spatial color histogram. The experimental results show that pedestrians can transfer moving objects and track between two cameras,and the tracking accuracy is significantly higher than that of single camera using SIFT algorithm directly.
In order to discover,warn,prevent and control the suspicious people and their behavior entering the subway station in time,the key is to realize the moving target transmission between multiple cameras. Therefore, the access channel of subway station is selected as this study background,and the monitoring image of two adjacent cameras installed in the access channel is taken as the research object. Vi Be algorithm is used to build the background detection model of moving target in two cameras with fixed and non-overlapping detection horizons. When pedestrians appear,the moving target template can be established to implement feature matching from the four aspects of pedestrian movement direction,height-width ratio,walking speed and spatial color histogram. The experimental results show that pedestrians can transfer moving objects and track between two cameras,and the tracking accuracy is significantly higher than that of single camera using SIFT algorithm directly.
引文
[1]黄新,黄柏华,熊显名.基于改进ORB算法无重叠视野域的目标交接[J].液晶与显示,2016(8):810-817.
[2]申明军,欧阳宁,莫建文,等.多摄像机环境下的目标跟踪[J].现代电子技术,2009(12):61-63,70.
[3]阳小燕,任家富,刘长江.基于视野分界线的多摄像机运动目标跟踪算法[J].计算机工程与应用,2016,52(9):208-212,238.
[4]张益男,袁杰.基于场边界线的多摄像机智能监控系统研究[J].现代电子技术,2015,38(21):6-10.
[5]汪鑫,刘嘉敏,李敏,等.基于SIFT特征匹配算法的目标跟踪及视频采集与传输研究[J].重庆理工大学学报(自然科学),2014,28(11):89-93.
[6]袁康,魏大鹏,赵从梅,等.一种基于SVM的核相关跟踪算法[J].传感器与微系统,2018,37(5):138-143.
[7]江少杰,宁纪锋,李云松.加权间隔结构化支持向量机目标跟踪算法[J].中国图象图形学报,2017,22(9):1261-1269.
[8]衡林,朱秀昌.无重叠视域的多摄像机之间的目标匹配[J].计算机工程与应用,2014(8):188-193.
[9]米向荣,曹建芳.基于局部敏感直方图的鲁棒目标跟踪[J].计算机工程与设计,2018,39(11):3526-3530,3548.
[10]冷建伟,李鹏.基于自适应特征分布更新的压缩跟踪算法[J].计算机工程,2018,44(2):264-270.
[11] DAVID G L. Distinctive Image Features from Scale Invariant Keypoints[J]. International Journal of Computer Vision,2004,60(2):91-110.
[12]顾苏杭,戎海龙,马正华.融合SIFT特征和HMM的运动目标识别与跟踪算法[J].科技通报,2017,33(11):210-215.
[13] COLLINS R T. Mean shift blob tracking through scale space[C]//Proceedings of the Conference on Computer Vision and Pattern Recognition. USA:IEEE,2003.
[14]郑涛,郝行猛,陈梅.基于特征匹配预估的Mean Shift运动检测与跟踪[J].传感器与微系统,2018,37(7):135-137,141.
[15]欧阳宁,刘钊.重叠视野域内多摄像机目标交换算法[J].激光与红外,2009(9):995-998.
[16]张正本,马同伟.基于卡尔曼一致滤波和分布式协作网络控制的多目标跟踪[J].电子测量与仪器学报,2018,32(3):127-134.
[17]陈智,柳培忠,骆炎民,等.自适应特征融合的多尺度相关滤波目标跟踪算法[J].计算机辅助设计与图形学学报,2018,30(11):2063-2073.
[18]庞岳峰,李春锋,牛攀峰,等.基于理论弹道速度匹配的飞行器测控记忆跟踪算法[J].电讯技术,2016,56(4):443-447.
[19]陈梦竹,岑翼刚,许家佗,等.基于图像处理的望诊面色自动识别研究[J].中国中医药信息杂志,2018,25(12):97-101.
[20] BARNICH O,VAN D M. Vibe:A universal background subtraction algorithm for video sequences[J]. IEEE Transactions on Image Processing,2011,20(6):1709-1724.
[21]王民,石新源,王稚慧,等.基于Vibe和时空上下文的运动手势跟踪算法[J].液晶与显示,2018,33(1):92-98.
[2]申明军,欧阳宁,莫建文,等.多摄像机环境下的目标跟踪[J].现代电子技术,2009(12):61-63,70.
[3]阳小燕,任家富,刘长江.基于视野分界线的多摄像机运动目标跟踪算法[J].计算机工程与应用,2016,52(9):208-212,238.
[4]张益男,袁杰.基于场边界线的多摄像机智能监控系统研究[J].现代电子技术,2015,38(21):6-10.
[5]汪鑫,刘嘉敏,李敏,等.基于SIFT特征匹配算法的目标跟踪及视频采集与传输研究[J].重庆理工大学学报(自然科学),2014,28(11):89-93.
[6]袁康,魏大鹏,赵从梅,等.一种基于SVM的核相关跟踪算法[J].传感器与微系统,2018,37(5):138-143.
[7]江少杰,宁纪锋,李云松.加权间隔结构化支持向量机目标跟踪算法[J].中国图象图形学报,2017,22(9):1261-1269.
[8]衡林,朱秀昌.无重叠视域的多摄像机之间的目标匹配[J].计算机工程与应用,2014(8):188-193.
[9]米向荣,曹建芳.基于局部敏感直方图的鲁棒目标跟踪[J].计算机工程与设计,2018,39(11):3526-3530,3548.
[10]冷建伟,李鹏.基于自适应特征分布更新的压缩跟踪算法[J].计算机工程,2018,44(2):264-270.
[11] DAVID G L. Distinctive Image Features from Scale Invariant Keypoints[J]. International Journal of Computer Vision,2004,60(2):91-110.
[12]顾苏杭,戎海龙,马正华.融合SIFT特征和HMM的运动目标识别与跟踪算法[J].科技通报,2017,33(11):210-215.
[13] COLLINS R T. Mean shift blob tracking through scale space[C]//Proceedings of the Conference on Computer Vision and Pattern Recognition. USA:IEEE,2003.
[14]郑涛,郝行猛,陈梅.基于特征匹配预估的Mean Shift运动检测与跟踪[J].传感器与微系统,2018,37(7):135-137,141.
[15]欧阳宁,刘钊.重叠视野域内多摄像机目标交换算法[J].激光与红外,2009(9):995-998.
[16]张正本,马同伟.基于卡尔曼一致滤波和分布式协作网络控制的多目标跟踪[J].电子测量与仪器学报,2018,32(3):127-134.
[17]陈智,柳培忠,骆炎民,等.自适应特征融合的多尺度相关滤波目标跟踪算法[J].计算机辅助设计与图形学学报,2018,30(11):2063-2073.
[18]庞岳峰,李春锋,牛攀峰,等.基于理论弹道速度匹配的飞行器测控记忆跟踪算法[J].电讯技术,2016,56(4):443-447.
[19]陈梦竹,岑翼刚,许家佗,等.基于图像处理的望诊面色自动识别研究[J].中国中医药信息杂志,2018,25(12):97-101.
[20] BARNICH O,VAN D M. Vibe:A universal background subtraction algorithm for video sequences[J]. IEEE Transactions on Image Processing,2011,20(6):1709-1724.
[21]王民,石新源,王稚慧,等.基于Vibe和时空上下文的运动手势跟踪算法[J].液晶与显示,2018,33(1):92-98.