基于序列复杂度的空中红外目标跟踪算法评估

Aerial infrared target tracking algorithm evaluation based on complexity of sequence

近年来,基于深度学习的红外空中目标跟踪算法不断涌现,如何对其性能进行评估已经成为一个亟待解决的问题。利用单帧图像混淆度和遮隐度对图像复杂度进行计算,并结合目标运动复杂度,建立了融合图像复杂度和运动复杂度的序列复杂度计算模型。构建了包含420个序列的红外序列样本库,利用序列复杂度对样本库测试结果进行加权评分,提出了一种新的红外空中目标跟踪算法性能评估方法。实验结果表明,所提出的评估方法能全面评估算法在不同态势下的性能。

In recent years,aerial infrared target tracking algorithms based on deep learning keep emerging.However,how to evaluate its performance has become an urgent problem to be solved.The complexity of the image is calculated by using the degree of the target being confused and the degree of the target being shielded of the single frame.The computing model of complexity of the sequence is designed based on the complexity of the image and the complexity of movement.Then,an infrared sequence sample database containing 420 sequences is constructed.A new algorithm evaluation method is proposed,in which the test results on the database are weighted through the complexity of the sequence.Experimental results show that it can comprehensively evaluate the performance of the algorithm in different states.