一种自适应双门限与KLT融合的目标跟踪方法  

Moving Target Tracking Method Based on Adaptive Double-threshold and KLT

作　　者：高捷 GAO Jie(Ministry of Communications,Navigation and Surveillance,Central-South Regional Air Traffic Management Bureau of China Civilaviation Administration,Guangzhou 510405,China)

机构地区：[1]民航中南空管局通信导航监视部,广州510405

出　　处：《小型微型计算机系统》2020年第5期946-951,共6页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(U1733111,U1833101)资助;四川省科技计划项目(2017JZ0007)资助;民航空管科技项目(2018XZ-214)资助.

摘　　要：为解决视频监控系统中目标跟踪过程出现的漂移问题,提出一种自适应双门限与KLT融合的目标跟踪方法.该方法首先采用自适应双门限运动目标检测获得高门限和低门限二值图像信息,同时将双门限二值图像信息与KLT目标特征点跟踪进行融合处理.如果KLT找到的特征点大部分在双门限目标的轮廓内,标记该轮廓为目标轮廓;否则,使用KLT特征点来确定目标的轮廓,最终实现场面复杂监视下运动目标的持续、准确跟踪.利用不同监控场景下跟踪多类目标体做实验,验证了本文方法不仅能够持续、准确地跟踪运动目标,而且可适用于场面雾天低能见度条件.To effectively solve the drifting problem of moving target tracking in video surveillance system,an moving target tracking method based on adaptive double-threshold and KLT is proposed.Firstly,adaptive double-threshold moving target detection is used to obtain the binary image information of high-threshold and low-threshold.At the same time,double-threshold binary image information is fused with target feature point tracking of KLT.If most of the feature points found by KLT are in the contour of double-threshold target,the contour is marked as the area of real target;otherwise,feature points of KLT are used to determine the contour of the target.Finally,the continuous and accurate tracking of moving targets is realized in complex video surveillance scenes.Experiments on multiple targets in different surveillance scene show that the proposed method can not only track moving targets continuously and accurately,but also be applied to foggy low visibility conditions.

关 键 词：视频监控 KLT 自适应双门限 目标跟踪 特征点 目标轮廓 

分 类 号：TP317[自动化与计算机技术—计算机软件与理论]