基于改进核相关滤波器的PTZ摄像机控制方法

doi:10.13973/j.cnki.robot.2016.0420

摘要
图/表
参考文献
相关文章 (15)

全文: PDF (720 KB) HTML ( ) 输出: BibTeX \| EndNote (RIS)
引用本文:
潘振福, 朱永利, 周国亮. 基于改进核相关滤波器的PTZ摄像机控制方法[J]. 机器人, 2016, 38(4): 420-427.DOI: 10.13973/j.cnki.robot.2016.0420. PAN Zhenfu, ZHU Yongli, ZHOU Guoliang. The PTZ Camera Control Method Based on the Improved Kernelized Correlation Filter. ROBOT, 2016, 38(4): 420-427. DOI: 10.13973/j.cnki.robot.2016.0420.

摘要针对传统的PTZ（pan-tilt-zoom）摄像机控制方法是依靠人工操作，无法连续、实时跟踪动态目标，且跟踪目标准确度低等缺陷，提出了一种基于改进的核相关滤波器（KCF）目标跟踪算法的PTZ摄像机控制方法．首先，对传统的KCF目标跟踪算法做了运动状态估计和尺度估计方面的改进．在目标运动状态估计中，将粒子滤波框架与传统KCF算法相结合，估算出运动目标的位置．这种基于概率的运动状态估计方法可以获得更加稳定的目标信号并同时减少背景干扰信息的引入，从而可以在复杂场景下具有更强的抗干扰性．目标尺度估计中采用相关滤波器在尺度金字塔中估算目标的尺度，使算法对尺度变化的运动目标具有更强的适应能力．其次，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．最后，将改进KCF算法与其他跟踪算法在Benchmark数据集中做对比实验，验证改进算法的鲁棒性与有效性．将算法应用于PTZ摄像机的控制中，并用C++语言实现了改进KCF算法控制PTZ摄像机上位机，实验结果表明该PTZ摄像机控制方法能准确跟踪被遮挡目标，使其稳定地呈现在取景框中．

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	潘振福
	朱永利
	周国亮

关键词 ：目标跟踪, 云台摄像机, 相关滤波器, 粒子滤波, 运动状态估计, 尺度估计

Abstract：Traditional PTZ (pan-tilt-zoom) camera control methods rely on manual operation. By those methods, dynamic objects can't be tracked continuously in real time, and the accuracy of object tracking is low. To solve those problems, a PTZ camera control method based on an improved KCF (kernelized correlation filter) object tracking algorithm is proposed. Firstly, the traditional KCF object tracking algorithm is improved in terms of motion state estimation and scale estimation. In target motion state estimation, the particle filter framework is combined with the traditional KCF algorithm to estimate the position of the moving target. With the motion state estimation method based on probability, stabler target signals can be obtained, and background interference information is reduced, thereby stronger anti-jamming ability is accomplished in complex scenarios. In target scale estimation, correlation filter is applied to estimating the scale of the target in scale pyramid, which improves the algorithm adaptability to the scale changes of moving targets. Secondly, the PTZ camera is controlled with the PELCO_D protocol according to the information of tracking results in order to keep the target within the viewfinder. Finally, the comparison experiments are carried out between the improved KCF algorithm and the other tracking algorithms using the Benchmark data sets in order to verify the effectiveness and robustness of the improved algorithm. The algorithm is applied to the PTZ camera control, and the PC system of PTZ camera is controlled by the improved KCF algorithm with C++ language. The experiment results show that the PTZ camera control method can track the obscured target accurately and keep it in the viewfinder stably.

Key words： object tracking PTZ (pan-tilt-zoom) camera correlation filter particle filtering motion state estimation scale estimation

收稿日期: 2016-04-05 录用日期: 2016-06-17

TP391

基金资助:河北省自然科学基金（F2014502069）

通讯作者: 潘振福，panzhenfu20@126.com E-mail: panzhenfu20@126.com

作者简介: 潘振福（1989-），男，硕士生．研究领域：计算机视觉，人工智能．
朱永利（1963-），男，博士，教授，博士生导师．研究领域：人工智能，机器学习．
周国亮（1978-），博士，副教授．研究领域：人工智能及应用．

链接本文:

https://robot.sia.cn/CN/10.13973/j.cnki.robot.2016.0420 或 https://robot.sia.cn/CN/Y2016/V38/I4/420

[1] 陈双叶,王善喜.PTZ 摄像机跟踪运动目标的智能控制算法的研究[J]. 计算机科学, 2015, 42(11A):135-139.Chen S Y, Wang S X. The study of the intelligent control algorithm for PTZ camera to track moving targets[J]. Computer Science, 2015, 42(11A): 135-139.
[2] Wang S, Tian Y, Xu Y P. Automatic control of PTZ camera based on object detection and scene partition[C]//IEEE International Conference on Signal Processing, Communications and Computing. Piscataway, USA: IEEE, 2015: 1-6.
[3] Dong E Z, Yan S X, Tong J G, et al. Design and implementation of a moving object tracking system[C]//IEEE International Conference on Mechatronics and Automation. Piscataway, USA: IEEE, 2015: 274-279.
[4] Kristan M, Pflugfelder R, Leonardis A, et al. The visual object tracking VOT2014 challenge results[M]//Lecture Notes in Computer Science, vol.8926. Berlin, Germany: Springer-Verlag, 2015: 191-217.
[5] Danelljan M, Khan F S, Felsberg M, et al. Adaptive color attributes for real-time visual tracking[C]//IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, USA: IEEE, 2014: 1090-1097.
[6] Henriques J F, Rui C, Martins P, et al. Exploiting the circulant structure of tracking-by-detection with kernels[M]//Lecture Notes in Computer Science, vol.7575. Berlin, Germany: Springer-Verlag, 2012: 702-715.
[7] Henriques J F, Caseiro R, Martins P, et al. High-speed tracking with kernelized correlation filters[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014, 37(3): 583-596.
[8] Bolme D S, Beveridge J R, Draper B, et al. Visual object tracking using adaptive correlation filters[C]//IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, USA: IEEE, 2010: 2544-2550.
[9] Danelljan M, Häger G, Khan F S, et al. Accurate scale estimation for robust visual tracking[C]//British Machine Vision Conference. 2014: 1-11.
[10] Muller K R, Mika S, Ratsch G, et al. An introduction to kernel-based learning algorithms[J]. IEEE Transactions on Neural Networks, 2001, 12(2): 181-201.

[11] Schölkopf B, Smola A J. Learning with kernels: Support vector machines, regularization, optimization, and beyond[M]. Boston, USA: MIT Press, 2001.
[12] Rifkin R, Yeo G, Poggio T. Regularized least-squares classification[M]//Suykens J A K, Horvath I, Basu S, et al. Advances in learning theory: Methods, model and applications, vol.190. Amsterdam, Netherland: IOS Press, 2003: 131-154.
[13] Gray R M. Toeplitz and circulant matrices: A review[J]. Foundations and Trends^® in Communications and Information Theory, 2006, 2(3): 155-239.
[14] 王法胜,赵清杰.一种用于解决非线性滤波问题的新型粒子滤波算法[J]. 计算机学报,2008,31(2):346-352.Wang F S, Zhao Q J. A new particle filter for nonlinear filtering problems[J]. Chinese Journal of Computers, 2008, 31(2): 346-352.
[15] Wu Y, Lim J, Yang M H. Visual tracker Benchmark[DB/OL]. [2013-10-12]. http://www.visual-tracking.net.
[16] Jia X, Lu H, Yang M H. Visual tracking via adaptive structural local sparse appearance model[C]//IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, USA: IEEE, 2012: 1822-1829.
[17] Wu Y, Lim J, Yang M H. Online object tracking: A benchmark[C]//IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, USA: IEEE, 2013: 2411-2418.
[18] Babenko B, Yang M H, Belongie S. Robust object tracking with online multiple instance learning[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(8): 1619-1632.