摘要
为了有效地实现复杂环境下机器人运动目标跟踪,提出了一种结合卡尔曼滤波和均值漂移的目标跟踪算法。该算法首先通过帧间差法在复杂背景中获取目标模型,以机器人自身一个周期的运动作为卡尔曼滤波器的输入量,以卡尔曼滤波器的估计值作为均值漂移算法的启动点,再利用均值漂移算法得到最终目标位置,最后通过目标遮挡判定来解决遮挡问题。实验表明:该算法能实现在复杂背景下对运动目标的稳定、准确的跟踪,对目标的遮挡有很好的鲁棒性。
A moving target tracking algorithm based on mean-shift algorithm and Kalman filter is proposed to effectively implement target tracking of mobile robot under complex environments.Through frame difference method,the target model is obtained under complex background,with the motion of robot as the input quantity of Kalman filter,the estimation value of Kalman filter as the starting position of the mean-shift algorithm and the mean-shift algorithm is used to obtain final target location.The occlusion judgment is used to solve occlusion problem.Experimental results show that the algorithm can track the moving target stably and accurately under complex environment and has good robustness to occlusion.
出处
《传感器与微系统》
CSCD
北大核心
2011年第6期112-115,共4页
Transducer and Microsystem Technologies