基于相关滤波的舰船目标抗干扰跟踪算法  

Anti-Jamming Tracking Algorithm for Ship Target Based on Correlation Filtering

作　　者：邱千钧 周鹏耀 高欣 张芳[4] 吕梅柏[2] Qiu Qianjun;Zhou Pengyao;Gao Xin;Zhang Fang;LüMeibo(The Second Military Representative Office of the Naval Equipment Department in Beijing,Beijing 100000,China;Northwestern Polytechnical University,Xi’an 710072,China;Military Industry Confidentiality Qualification Review and Certification Center,Beijing 100089,China;Xi’an Institute of Applied Optics,Xi’an 710065,China)

机构地区：[1]海军装备部驻北京地区第二军事代表室,北京100000 [2]西北工业大学,西安710072 [3]军工保密资格审查认证中心,北京100089 [4]西安应用光学研究所,西安710065

出　　处：《航空兵器》2023年第6期123-129,共7页Aero Weaponry

基　　金：基础加强计划重点基础研究项目(2020-JCJQ-ZD-076-00);哲学社会科学基金项目(21AGL030)。

摘　　要：针对舰船类目标跟踪中面临的形态变化和尺度缩放导致的跟踪漂移或丢失问题,本文设计了一种基于多特征融合和阈值选取的抗形态变化和尺度变换干扰目标跟踪算法。在抗形态变化模块中,设计了一种多特征加权融合方法。通过将方向梯度直方图、局部二值模式、颜色矩按颜色矩特征识别贡献率进行自适应加权融合,加强了舰船类目标重要部位特征提取能力,提高了算法跟踪过程的鲁棒性;在抗尺度变换干扰模块中,设计了一种多分辨率目标框共同搜索、最大响应峰确定目标位置的方法,解决了舰船类目标尺度缩放导致的跟踪框鲁棒性低的问题。实验结果表明,本文算法具有更出色的跟踪性能,其在OTB数据集上的精确度为93.6%,成功率达到70.1%,优于其他相关算法。For solving the problem of tracking drift or loss caused by morphological changes and scale scaling in ship target tracking,this paper designs a target tracking algorithm based on multi feature fusion and threshold selection to resist morphological changes and scale transformation interference.In the anti morphological change module,a multi feature weighted fusion method is designed.Through the adaptive weighted fusion of histogram of oriented gradients(HOG),local bninary patterns(LBP)and color names(CN)by the contribution rate of color moment feature recognition,the feature extraction ability of important parts of ship targets is strengthened,and the robustness of the proposed algorithm in the tracking process is improved.In the anti scale transformation interference module,a method with multi resolution target box joint search method and determination of target position by the maximum response peak is designed to solve the problem of low robustness of the tracking box caused by the scaling of ship targets.The experimental results show that the proposed algorithm has better tracking performance,with an accuracy of 93.6%and a success rate of 70.1%on the OTB dataset.This method is superior to other related algorithms.

关 键 词：特征融合 阈值选取 目标跟踪 相关滤波 尺度缩放 方向梯度直方图 抗干扰 

分 类 号：TJ760[兵器科学与技术—武器系统与运用工程]