基于层次相关滤波的无人机视觉目标跟踪  

UAV Visual Object Tracking Method Based On Hierarchical Correlation Filter

作　　者：刘阳 Liu Yang(College of Information Engineering,Hunan Industry Polytechnic,Changsha 410208,China)

机构地区：[1]湖南工业职业技术学院信息工程学院,长沙410208

出　　处：《办公自动化》2025年第4期103-107,共5页Office Informatization

基　　金：2023年湖南工业职业技术学院校级课题“基于层级相关滤波和注意力机制的无人机视觉目标跟踪方法研究”(项目编号:GYXJKT202315)。

摘　　要：针对无人机视觉目标跟踪中存在的挑战场景,文章提出一种基于层次相关滤波的跟踪算法,该算法利用预训练的VGGNet-19网络提取目标的多层卷积特征,构建多模型并行跟踪框架,并结合模型置信度评估机制,实现对滤波器模型的自适应更新。在UAV123数据集上的实验结果表明,该算法在精度和成功率上分别达到68.6%和61.0%,能有效应对目标遮挡、形变等挑战,实现对无人机视野下目标的精准和快速跟踪。To address the challenges in UAV visual object tracking,this paper proposes a tracking algorithm based on hierarchical correlation filters.The algorithm utilizes a pre-trained VGGNet-19 network to extract multi-layer convolutional features of the target,constructs a multi-model parallel tracking framework,and integrates a model confidence evaluation mechanism for adaptive updating of the filter models.Experimental results on the UAV123 dataset demonstrate that the proposed method achieves an accuracy of 68.6% and a success rate of 61.0%,effectively handling challenges such as target occlusion and deformation,and enabling precise and rapid tracking of objects within the UAV's field of view.

关 键 词：无人机 计算机视觉 相关滤波 多层卷积特征 自适应更新 

分 类 号：V279[航空宇航科学与技术—飞行器设计] TP391.41[自动化与计算机技术—计算机应用技术]