夜间边境非法越界热成像智能监测方法
|
摘要
国家边界线夜间非法越境监测和报警是一个重要的研究课题,目前存在的监控系统智能性低、误检率高、计算时间慢。为了实现智能监测,提出了一种热成像智能监测方法。首先使用高斯背景建模提取前景目标,使用HOG+SVM的方法对前景目标进行越界人检测;为了消除动物的干扰,利用STC算法进一步跟踪并分析目标速度和运动轨迹。实验结果表明,设计的方法提高了越界人检测的准确率,检测时间提高到60 ms左右,满足了现场应用。
Detecting illegal Person crossing national boundary line at night is an important research topic. The current monitoring system has disadvantage of low intelligence,high false detection rate and slow calculation time and so on.In order to achieve intelligent monitoring,a thermography intelligent illegal person monitoring method is proposed.Firstly,the foreground object is extracted using Gaussian background modeling in order to decrease the processed image size. The HOG and SVM are combined to detect the running target. Then,in order to eliminate the false alarm caused by animal interference,the STC algorithm is used to get the speed and trajectory of detected person or animal and judge illegal cross-border person according to running trajectory. The experimental results show that this method not only improves the illegal person detection accuracy,but also every frame average detection time is only less than60 ms which satisfies the actual application requirement.
Detecting illegal Person crossing national boundary line at night is an important research topic. The current monitoring system has disadvantage of low intelligence,high false detection rate and slow calculation time and so on.In order to achieve intelligent monitoring,a thermography intelligent illegal person monitoring method is proposed.Firstly,the foreground object is extracted using Gaussian background modeling in order to decrease the processed image size. The HOG and SVM are combined to detect the running target. Then,in order to eliminate the false alarm caused by animal interference,the STC algorithm is used to get the speed and trajectory of detected person or animal and judge illegal cross-border person according to running trajectory. The experimental results show that this method not only improves the illegal person detection accuracy,but also every frame average detection time is only less than60 ms which satisfies the actual application requirement.
引文
[1]王斌,刘洋,唐胜.融合多模型和帧间信息的行人检测算法[J].计算机辅助设计与图形学学报,2017,29(3):444-449.
[2]杨颖,黄晓峰.一种复杂背景下多尺度多视角的人体检测方法[J].计算机应用与软件,2016,33(8):329-333.
[3]向涛,李涛,李旭冬,等.基于随机森林的层次行人检测算法[J].计算机应用研究,2015,32(7):2196-2199.
[4]胡庆新,张春阳,方静.基于多特征的行人检测算法[J].计算机应用研究,2014,31(10):3161-3164.
[5] Liu H,Xu T,Wang X,et al. Related HOG Features for Human Detection Using Cascaded Adaboost and SVM Classifiers[M].Advances in Multimedia Modeling. Springer Berlin Heidelberg,2013(7733):345-355.
[6] Stauffer C,Grimson W E L. Adaptive Background Mixture Models for Real-Time Tracking[C]//cvpr. IEEE Computer Society,1999:2-2246.
[7] Bertozzi M,Broggi A,Rose M D,et al. A Pedestrian Detector Using Histograms of Oriented Gradients and a Support Vector Machine Classifier[C]//Intelligent Transportation Systems ConferenceItsc.IEEE,2007:143-148.
[8]胡正平,杨建秀. HOG特征混合模型结合隐SVM的感兴趣目标检测定位算法[C]//全国信号处理学术年会. 2011(3):115-127.
[9] Song H,Xiao B,Qinzhen H U,et al. Integrating Local Binary Patterns into Normalized Moment of Inertia for Updating Tracking Templates[J]. Chinese Journal of Electronics,2016,25(4):706-710.
[10] Song H,Zhou G,Zhao J,et al. A Practical and Effective Intelligent Traffic Counting Method[J]. ICIC Express Letters,Part B:Applications,2017,8:1095-1101.
[11] Zhang K,Zhang L,Yang M H,et al. Fast Tracking via Spatio-Temporal Context Learning[J]. Computer Science,2013.
[12]孙红星,王嘉艺,刘大鹏.基于运动轨迹的徘徊行为检测[J].辽宁科技大学学报,2014,37(4):387-392.
[2]杨颖,黄晓峰.一种复杂背景下多尺度多视角的人体检测方法[J].计算机应用与软件,2016,33(8):329-333.
[3]向涛,李涛,李旭冬,等.基于随机森林的层次行人检测算法[J].计算机应用研究,2015,32(7):2196-2199.
[4]胡庆新,张春阳,方静.基于多特征的行人检测算法[J].计算机应用研究,2014,31(10):3161-3164.
[5] Liu H,Xu T,Wang X,et al. Related HOG Features for Human Detection Using Cascaded Adaboost and SVM Classifiers[M].Advances in Multimedia Modeling. Springer Berlin Heidelberg,2013(7733):345-355.
[6] Stauffer C,Grimson W E L. Adaptive Background Mixture Models for Real-Time Tracking[C]//cvpr. IEEE Computer Society,1999:2-2246.
[7] Bertozzi M,Broggi A,Rose M D,et al. A Pedestrian Detector Using Histograms of Oriented Gradients and a Support Vector Machine Classifier[C]//Intelligent Transportation Systems ConferenceItsc.IEEE,2007:143-148.
[8]胡正平,杨建秀. HOG特征混合模型结合隐SVM的感兴趣目标检测定位算法[C]//全国信号处理学术年会. 2011(3):115-127.
[9] Song H,Xiao B,Qinzhen H U,et al. Integrating Local Binary Patterns into Normalized Moment of Inertia for Updating Tracking Templates[J]. Chinese Journal of Electronics,2016,25(4):706-710.
[10] Song H,Zhou G,Zhao J,et al. A Practical and Effective Intelligent Traffic Counting Method[J]. ICIC Express Letters,Part B:Applications,2017,8:1095-1101.
[11] Zhang K,Zhang L,Yang M H,et al. Fast Tracking via Spatio-Temporal Context Learning[J]. Computer Science,2013.
[12]孙红星,王嘉艺,刘大鹏.基于运动轨迹的徘徊行为检测[J].辽宁科技大学学报,2014,37(4):387-392.