摘要
针对复杂场景中存在受目标相似物干扰造成的跟踪错误问题,改进了基于颜色直方图概率分布的跟踪算法,首先利用卡尔曼滤波器预估目标状态所在区域,可避免目标被遮挡时Cam-shift算法陷入局部最大值,以及目标速度过快时导致跟踪失败的问题。然后利用马尔科夫模型加入方向预测器,在背景干扰下实现颜色特征离散时的目标状态估计,增强检测器对相似目标的辨识能力。通过对多组场景下的目标跟踪实验,结果表明,改进后的跟踪算法提高了跟踪准确度和目标的检测速度,并且很大程度上提高了复杂背景中辨识相似目标的准确度。
Aiming at the problem of tracking errors is caused by interference with objects similar to object in complex scenes,the tracking algorithm is improved based on color histogram probability distribution.In the first place,Kalman filter is used to estimate the target state area in order to avoid the problem of Cam-shift algorithm's falling into local maximum when the target is blocked and the problem of target tracking failure when the target moves too quickly.Then the direction predictor based on the Markov model is added to realize the target state estimation under the condition of background color interference and to enhance the ability of the detector to identify similar objects.The experimental results show that the new algorithm can improve the accuracy of target tracking as well as the target testing speed,and the accuracy of identifying similar targets in complex scenes in a great sense.
出处
《湖北工程学院学报》
2017年第6期102-106,共5页
Journal of Hubei Engineering University
基金
武汉工程大学研究生创新基金项目(CX2016071)
关键词
目标跟踪
卡尔曼滤波
遮挡
处理速度
目标预估
target tracking
Kalmanfilter
occlusion
processing speed
target estimation