跟踪系统持续稳定跟踪目标的粒子滤波  

Research on Armored Fire Control Video Long-term Tracking System Based on KCF and Particle Filtering Algorithm

作　　者：范玲瑜[1] 王世明[1] 张超[1] 丁玲[1] 胡冠华[1] FAN Ling-yu;WANG Shi-ming;ZHANG Chao;DING Ling;HU Guan-hua(North Automatic Control Technology Institute,Taiyuan 030006,Chin)

机构地区：[1]北方自动控制技术研究所,太原030006

出　　处：《火力与指挥控制》2018年第6期136-140,共5页Fire Control & Command Control

摘　　要：在现代战争中,视频跟踪技术是实现精确火力打击的关键技术,是获取信息对抗胜利从而夺取战争主动权、实现发现即摧毁的重要保证。由于战场地理环境复杂,经常存在跟踪目标被遮挡而丢失的情况,因此,稳定的长时跟踪算法必不可少。针对现在战场复杂环境下目标跟踪遇到的问题,提出了基于KCF(Kernelized Correlation Filters)和粒子滤波相融合的跟踪算法,相比较传统算法其能够在复杂环境下完成长时、稳定、准确的跟踪。通过设计的视频跟踪系统的硬件平台实验结果显示,该算法可实现目标的稳定跟踪以及目标被遮挡丢失之后的重新检测,能够满足复杂战场环境下目标的长时稳定跟踪的需求。Video tr acking technology is the key technology to realize precision fire attack in modern warfare. It is an important guarantee to achieve success for the information confrontation,and it is a prerequisite for seizing the initiative of war and realizing discovery and destruction. As the battlefield geographical environment is complex,there is often a case where the tracking target is obscured and lost,so a stable long-term tracking algorithm is essential. This paper proposes a tracking algorithm based on KCF（Kernelized Correlation Filters）and Particle Filtering,which can achieve longterm,stable and accurate tracking in complex environment. The experimental results show that the algorithm can achieve stable tracking of the target and re-detection when the target is obscured,which can meet the requirement of long-time stable tracking in complex battlefield environment.

关 键 词：视频跟踪 KCF 粒子滤波 丢失重检 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] TJ02[自动化与计算机技术—控制科学与工程]