基于目标信息量变化的自调整跟踪算法

针对视频跟踪中的传统Mean-Shift方法不能有效跟踪尺度有明显变化的目标这一问题,将图像信息量度量理论引入到了算法之中,提出了一种自适应窗宽的改进Mean-Shift跟踪算法。该算法利用均值漂移矢量对目标位置进行预测,然后结合目标信息量的变化来自适应的调整跟踪窗的尺度,从而快速稳定的对目标进行定位跟踪。实验结果表明改进算法能较好适应目标的尺度变化,跟踪效果良好。

Adaptive Bandwidth Tracking Algorithm Based On Object Information Change

The traditional Mean-Shift algorithm in video tracking can not effectively track the significant change in the target scale.To solve the problem,an adaptive bandwidth Mean-Shift tracking algorithm based on the image information measure theory is proposed. The improved algorithm using Mean-Shift vector preliminary predicts the target position,then combining with the information change of object achieves tracking windows’ automatic adjustment.Thus,it can rapidly to target positioning,and stable for tracking.The experimental results indicate that the improved algorithm can better adapt to the scale changes of target,and the tracking effect is great.