复杂背景下的红外小目标跟踪

Small Infrared Target Tracking in Complex Background

复杂背景下的红外小目标易受背景影响,且由于距离系统较远,能够提供的有效信息较少,这给目标的跟踪造成了困难。文章在传统KCF算法的基础上进行改进,将灰度信息与HOG特征进行加权融合,更好地表征目标;增加自适应尺度估计,弥补原算法不能改变尺度的缺陷;引入重检测机制,自适应地对模型进行更新,有效解决目标或背景发生变化时的跟踪漂移问题。实验结果表明:文中算法与原算法相比,跟踪精度与鲁棒性有所提升,能够实现复杂背景下的红外小目标的持续跟踪。

Small infrared targets in complex background are easily affected by the background,and because of the distance from the system,it provides less effective information,which make it much difficult to track the target.In this paper,the KCF algorithm is improved.The gray feature and HOG feature are fused to represent the target better.Adaptive scale estimation is added to make up for the defect that the KCF algorithm,but it cannot change the scale.The redetection mechanism is introduced to update the module,to solve the problem of target loss when the background or target changes.Experimental results show that,compared with the original algorithm,the proposed algorithm has improved tracking accuracy and robustness,which can realize tracking of infrared small targets under complex background.