基于速度矢量匹配的低特征破片群目标跟踪  

Low feature fragment group target tracking based on velocity vector matching

作　　者：李乾舞 赵建新 张玉荣 杜博军 刘泽庆 李天宇 LI Qian⁃wu;ZHAO Jian⁃xin;ZHANG Yu⁃rong;DU Bo⁃jun;LIU Ze⁃qing;LI Tian⁃yu(Artillery Engineering Department,Shijiazhuang Campus of Army Engineering University,Shijiazhuang 050003,China;Department of Aviation,Shijiazhuang Engineering Vocational College,Shijiazhuang 050061,China;Unit 63850 of PLA,Baicheng 137001,China;Unit 32306 of PLA,Pingdingshan 467000,China)

机构地区：[1]陆军工程大学石家庄校区火炮工程系,河北石家庄050003 [2]石家庄工程职业学院航空系,河北石家庄050061 [3]中国人民解放军63850部队,吉林白城137001 [4]中国人民解放军32306部队,河南平顶山467000

出　　处：《电子设计工程》2020年第1期59-64,共6页Electronic Design Engineering

摘　　要：静爆测试中破片具有体积小,速度快的特征,在高速摄像机图像序列中与背景对比度低,为了有效提取到破片目标,完成序列中的目标跟踪,提出基于速度矢量的低特征目标群的跟踪方法。方法利用前后帧时间间隔短速度矢量变换可忽略的特点,建立适合破片群运动特征的相关函数,结合新三步法完成对同名点的搜索。实验对某次静爆测试图像序列进行处理,成功提取到破片目标,并完成破片轨迹的模拟,召回率在79%,能够满足靶场测试的需要,为后期双目视觉匹配提供依据。In static explosion test,fragments have the characteristics of small size and fast speed. The contrast between fragments and background is low in high-speed camera image sequence. In order to extract fragments effectively and complete target tracking in sequence,a tracking method based on low feature target group of motion vector is proposed. Methods Based on the negligible feature of short speed vector transformation between front and back frames,a correlation function suitable for fragment group motion characteristics was established,and a new three-step method was used to complete the search of homonymous points. After processing the image sequence of a static explosion test,the fragment target is successfully extracted and the fragment trajectory is simulated. The recall rate is 79%,which can meet the needs of range test and provide the basis for binocular vision matching in the later stage.

关 键 词：多目标图像 速度矢量 高速摄像 点匹配 

分 类 号：TJ06[兵器科学与技术—兵器发射理论与技术]