基于改进的MeanShift算法的选煤厂人员目标跟踪方法

Personnel Tracking Method of Coal Preparation Plant Based on Improved MeanShift Algorithm

  • 摘要: 针对采用传统的MeanShift算法进行智能视频监控易受背景干扰而丢失目标的问题,提出了一种将MeanShift算法与卡尔曼滤波算法相结合的选煤厂人员目标跟踪方法。该方法首先通过运动检测方法分割出跟踪目标区域,然后通过卡尔曼滤波算法预测下一帧跟踪窗口的起点,在此基础上采用MeanShift算法跟踪目标区域;由于选煤厂环境较复杂,为了防止跟踪失败,采用跟踪与检测相结合的方法来进一步保证跟踪的鲁棒性。实验结果表明,该方法能很好地消除背景中相似颜色区域的影响,具有较好的跟踪效果。

     

    Abstract: In view of problem of object losing because that traditional MeanShift algorithm used in intelligent video monitoring is easy to be disturbed by background, the paper proposed a personnel tracking method of coal preparation plant combining with MeanShift algorithm and Kalman filtering algorithm. The method firstly segments object tracking region through motion detection method, and predicts starting point of next frame tracking window through Kalman filtering algorithm, then uses MeanShift algorithm to track object region on the basis. In order to prevent tracking failure because of complex environment of coal preparation plant, the method combines with tracking and detection to further ensure robustness of tracking. The experiment result showed that the method can eliminate influence of similar color region in background and has better tracking effect.

     

/

返回文章
返回