基于视觉目标跟踪的PTZ控制方法  

The method of controlling PTZ camera based on visual object tracking

作　　者：潘振福 许静[1] 刘成[1] 朱永利[1] Pan Zhenfu Xu Jing Liu Cheng Zhu Yongli(Department of Computer Science, North China Electric Power University, Baoding 071003, China)

机构地区：[1]华北电力大学计算机系,河北保定071003

出　　处：《电子技术应用》2017年第1期145-147,共3页Application of Electronic Technique

摘　　要：针对目前基于主动视觉的PTZ摄像机控制跟踪性能差,无法连续、实时跟踪动态目标,且跟踪目标的准确度低下等缺陷,提出了一种基于核相关视觉目标跟踪算法的云台摄像机控制方法。首先设计了云台摄像机系统的整体架构。视觉目标跟踪采用核相关目标跟踪方法,时效性很高,跟踪精确度也位列于目标跟踪领域的高等水平。根据跟踪结果信息,通过PELCO-D协议控制PTZ摄像机,始终保持目标在视频画面内。并用C++实现了KCF算法控制PTZ摄像机上位机,实验验证了该种PTZ控制方法的准确性、适用性及稳定性。For that currently the PTZ camera control based on active tracking is performing poor, not able to tack dynamic object continuously and real-timely, and low accuracy of tracking,a method of controlling PTZ camera based on kernelized correlation ob- ject tracking is proposed. Firstly, the overall architecture of PTZ camera system is designed. The KCF tracker used is efficient.And it＇s tracking accuracy is also ranked in higher level in the field of target tracking.Secondly,to make sure keeping the target in video images,PTZ camera is controlled through the PELCO_D protocol with the information of tracking results. The PC system of controlling PTZ camera is implemented by using Improved KCF algorithm with C＋＋. The accuracy, applicability and stability of this kind of PTZ control method are verified by the experiments.

关 键 词：目标跟踪 云台控制 相关滤波器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]