基于多域网络的红外目标跟踪算法

随着红外探测器性能的不断提升, 红外目标跟踪在智能安防等领域的应用越来越广泛。然而红外图像分辨率较低, 依赖其进行目标跟踪仍然是一个制约相关应用发展的难题。针对红外图像分辨率低的特点, 以多域网络为模型算法框架, 结合目标运动过程中的尺寸变化特点, 提出了一种基于多域网络的红外目标跟踪算法。为了评估算法性能, 分别在VOT-TIR2016数据库和AMCOM红外数据上进行了实验。实验结果表明, 目标尺度预测机制能够显著提高算法的跟踪精度。

Infrared target tracking algorithm based on multi-domain network

With the continuous improvement of infrared detector performance,infrared target tracking is more and more widely used in intelligent security and other fields.However,it is still a difficult problem to rely on infrared image for target tracking due to the low resolution of infrared image.In view of the low resolution of infrared image,combining with the characteristics of size change in the process of target movement,a multi-domain network based infrared target tracking algorithm is proposed,which is based on the multi-domain network.To assess the algorithm performance,experiments are conducted on the VOT-TIR2016 database and AMCOM infrared data,respectively.Experimental results show that the target scale prediction mechanism can significantly improve the tracking accuracy of the algorithm.