复杂背景下兼顾跟踪实时性和跟踪精度的目标跟踪技术研究  

Research on target tracking technology considering tracking real-time and tracking precision in complex background

作　　者：李庆生[1] 赵丽君[2] 张志锋[3] LI Qing-sheng;ZHAO Li-jun;ZHANG Zhi-feng(Safety Work Department of Chengde Petroleum College,Chengde 067000,China;Department of Electrical and Electronic Engineering,Chengde Petroleum College,Chengde 067000,China;Shenyang School of Electrical Engineering,University of Technology,Shenyang 110870,China)

机构地区：[1]承德石油高等专科学校安全工作处,河北承德067000 [2]承德石油高等专科学校电气与电子工程系,河北承德067000 [3]沈阳工业大学电气工程学院,辽宁沈阳110870

出　　处：《光电子．激光》2020年第2期117-124,共8页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金(61603263)资助项目。

摘　　要：由于将CamShift算法在复杂背景和操作条件下应用于视频跟踪,跟踪失败和目标损失的现象将非常容易发生。为了提高复杂环境条件下目标跟踪的精度及实时性,本论文提出了一种能够在复杂环境条件下及时对目标对象进行追踪的技术。以颜色、纹理、目标动作信息的全面特性为基础对CamShift算法作出整改完善,通过组合Kalman过滤器预评估目标对象的动作情况,在目标对象受到制约的情况下,使用运转前的目标对象预先信息,对目标对象物体的动作轨迹执行最小平方运算以及外穿推进,同时基于对象物体的位移情况进行定位信息的预测评估,以助于恢复目标的定位信息直到制约情况结束。经多次实验,相关统计数据表明,这一算法能够用于复杂情形的环境条件下,且当目标对象处于短期闭塞情况下依然能达成目标的连续稳定追踪,在性能上具备出色的实时性。Since the CamShift algorithm is applied to video tracking under complex background and operating conditions,tracking failure and target loss will easily occur.In order to improve the accuracy and real-time of target tracking under complex environmental conditions,this paper proposes a technique that can track target objects in time under complex environmental conditions.Based on the comprehensive characteristics of color,texture and target action information,the CamShift algorithm was rectified and improved,the Kalman filter was combined to pre-evaluate the motion of the target object.When the target object is restricted,the pre-information of the target object before operation is used,and the moving track of the target object is fitted with the least square method and pushed outward,and the positioning information is predicted and evaluated around the displacement of the target object.This will help to recover the positioning information of the target until the end of the constraint.After many experiments,relevant statistical datashow that this algorithm can be used in the environmental conditions of complex situations,and the target can still achieve continuous and stable tracking when the target is in a short-term occlusion situation,and has excellent real-time performance.

关 键 词：复杂情形 目标跟踪移动 CAMSHIFT算法 Kalman过滤器 预测评估 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]