摘要
提出一种基于改进高斯混合模型和卡尔曼滤波的车辆检测与跟踪方法.该方法在车辆检测阶段,为了解决传统高斯混合模型对运动目标速度变化自适应能力较差的问题,通过定义运动目标速率因子,给出一种模型学习率自适应更新策略,对传统高斯混合模型进行了改进,并用以实现车辆检测;在车辆跟踪阶段,通过建立一个适用于视频目标跟踪的卡尔曼滤波系统,并以车辆检测阶段输出的车辆质心为该卡尔曼滤波系统的量测值,实现了选定车辆目标的跟踪.实验结果表明,该方法车辆检测与跟踪效果良好,能满足实际交通监控系统的需求.
A vehicle detecting and tracking method was proposed by using an improved Gaussian mixture model（GMM）and Kalman filter（KF）.In the vehicle detecting stage,an improved GMM was presented to solve the problem that the traditional GMM has weak adaptability to the speed changes of moving target,in which a moving target speed factor was designed and used to adaptively update the learning rate of GMM.Using the improved GMM,a vehicle detecting scheme was also proposed.In the vehicle tracking stage,a KF system for video target tracking was designed firstly.Using the vehicle centroid,which is the output of the vehicle detecting scheme,as measurement of the KF system,the selected vehicle can be tracked effectively.Experiment results show that the proposed method can detect and track vehicle effectively and meet the demand of actual traffic monitoring system.
出处
《河南大学学报（自然科学版）》
CAS
2017年第6期693-698,共6页
Journal of Henan University:Natural Science
基金
国家自然科学基金项目(U1504621)
河南省科技发展计划项目(172102210185)
关键词
高斯混合模型
卡尔曼滤波
车辆检测
车辆跟踪
交通视频处理
Gaussian mixture model
Kalman filter
vehicle detecting
vehicle tracking
transport video processing
