基于线性核函数的高速核化相关跟踪算法

刘鑫昌; 冯路; 李继东; 马钟; 毕瑞星

doi:10.19304/J.ISSN1000-7180.2021.1348

基于线性核函数的高速核化相关跟踪算法

doi: 10.19304/J.ISSN1000-7180.2021.1348

1.
驻西安地区某军代室，陕西西安 710000
2.
西安微电子技术研究所，陕西西安 710054

基金项目:

中国航天科技集团有限公司发展战略研究课题 FZZLYJ-2020-013

详细信息

作者简介:
刘鑫昌男，(1979-)，博士，工程师.研究方向为计算机应用、合同监管

冯路男，(1978-)，高级工程师.研究方向为自动控制、合同监管

李继东男，(1976-)，工程师.研究方向为指挥自动化、合同监管

毕瑞星男，(1989-)，硕士，工程师.研究方向为计算机视觉

通讯作者:
马钟(通讯作者) 男，(1984-)，博士，研究员.研究方向为计算机视觉、机器学习，人工智能. E-mail：mazhong@mail.com

中图分类号: TP3
计量
- 文章访问数: 92
- HTML全文浏览量: 61
- PDF下载量: 8
- 被引次数: 0
出版历程
- 收稿日期: 2021-12-20
- 修回日期: 2022-03-08
- 网络出版日期: 2022-11-29

High-speed correlation tracking algorithm based on linear kernel function

1.
A military representative office stationed in Xi′an, Xi′an 710000, China
2.
Xi′an Microelectronics Technology Institute, Xi′an 710054, China

摘要

摘要:
现有的视觉目标跟踪研究主要集中在跟踪性能的提升，计算量普遍过大，难以在计算资源有限的嵌入式计算平台上实时运行，严重影响了跟踪算法的实际应用.对现有的跟踪算法进行了分析，提出了一种改进的高速核化相关跟踪算法.一方面，采用线性核函数解决相关运算中核函数计算量大的问题，另一方面，对算法流程进行优化，将多个傅里叶变换计算放在算法初始化阶段，从而避免在跟踪过程中进行运算量较大的傅里叶变换计算.综合上述措施，将原来的每次跟踪主循环需要计算十次傅里叶变换(FFT)减少到三次FFT.并通过定量实验分析验证，在跟踪性能基本不变的情况下，将速度提升到原来的4-5倍.提出的方法大幅降低了高性能跟踪算法的计算量，在计算性能有限的嵌入式计算平台上有着良好的应用前景.
- 目标跟踪 /
- 线性核函数 /
- 核化相关跟踪
Abstract:
The existing research on visual target tracking mainly focuses on the improvement of tracking performance. The amount of computation is generally too large to run in real-time on embedded computing platforms with limited resources, which seriously affects the practical application of tracking algorithms. This paper analyzes the existing tracking algorithms, and proposes an improved high-speed kernelized correlation tracking algorithm. On the one hand, the linear kernel function is used to solve the problem of a large amount of kernel function calculation in the correlation operation, on the other hand, the algorithm flow is optimized, and multiple Fourier transform calculations are placed in the algorithm initialization stage, so as to avoid the large amount of Fourier transform calculation in the tracking process. Combining the above measures, the original tracking main cycle needs to calculate ten times Fourier transform (FFT) to three times FFT. And through quantitative experimental analysis and verification, the speed of the proposed algorithm is increased to 4-5 times that of the original tracking algorithm, while the tracking performance is basically unchanged. The proposed method in this paper dramatically reduces the computational complexity of high-performance tracking algorithms and has a good application prospect on embedded computing platforms with limited computing performance.
- object tracking /
- liner kernel function /
- kernelized correlation tracking

HTML全文

图 1 核化相关跟踪算法跟踪基本流程

Figure 1. Basic flow of the correlation tracking algorithm

下载: 全尺寸图片幻灯片

图 2 修改后的算法流程图

Figure 2. Flow chart of the improved algorithm

下载: 全尺寸图片幻灯片

图 3 跟踪效果示例

Figure 3. Example of tracking results

下载: 全尺寸图片幻灯片

图 4 改进前后算法在测试数据集上的定量对比

Figure 4. Quantitative comparison of the improved algorithm on the test dataset

下载: 全尺寸图片幻灯片

表 1 改进后的kcf算法的定量实验结果

Table 1. Quantitative experimental results of the improvedkcf algorithm

图像数据来源	图像序列	精确性		时间消耗/ms		鲁棒性
图像数据来源	图像序列	原始KCF算法	改进后的KCF算法	原始KCF算法	改进后的KCF算法	原始KCF算法	改进后的KCF算法
vot dataset	road	0.571585	0.555519	12.421 9	2.54409	0	0
vot dataset	racing	0.311684	0.276357	11.925 9	2.17734	0	0
vividccd dataset	RedTeam	0.51354	0.456355	7.8886	3.66688	0	2
	EgTest01	0.488605	0.518223	13.975	3.06616	0	2
	EgTest02	0.436159	0.50454	12.5327	2.781	0	4
	EgTest03	0.538482	0.465848	12.7075	3.05701	1	3
	EgTest04	0.63546	0.34188	10.5922	2.37499	4	6
	EgTest05	0.371108	0.41711	12.9239	2.48624	4	9
vividirdataset	PkTest01	0.462492	0.409642	11.5117	3.30028	15	16
	PkTest02	0.288126	0.298407	11.6544	3.82546	4	5
	PkTest03	0.650305	0.487371	10.9915	3.36915	11	13
均值		0.4789	0.4301	11.7387	2.9681	3.5455	5.4545

下载: 导出CSV

参考文献(12)

[1]	张格森, 陈东生, 邱海韬, 等. 基于FPGA的高速图像跟踪系统设计[J]. 微电子学与计算机, 2017, 34(4): 13-16. DOI: 10.19304/j.cnki.issn1000-7180.2017.04.003. ZHANG G S, CHEN D S, QIU H T, et al. Design of high speed image tracking system based on FPGA[J]. Microelectronics & Computer, 2017, 34(4): 13-16. DOI: 10.19304/j.cnki.issn1000-7180.2017.04.003.
[2]	ZHU P F, WEN L Y, DU D W, et al. Detection and tracking meet drones challenge[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021. DOI: 10.1109/TPAMI.2021.3119563.
[3]	KRISTAN M, MATAS J, LEONARDIS A, et al. The ninth visual object tracking VOT2021 challenge results[C]//Proceedings of 2021 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW). Montreal: IEEE, 2021: 2711-2738. DOI: 10.1109/ICCVW54120.2021.00305.
[4]	程钰清, 贺占庄, 马钟, 等. 面向嵌入式FPGA的智能目标检测算法[J]. 微电子学与计算机, 2021, 38(6): 87-92. DOI: 10.19304/j.cnki.issn1000-7180.2021.06.015. CHENG Y Q, HE Z Z, MA Z, et al. Intelligent target detection algorithm for embedded FPGA[J]. Microelectronics & Computer, 2021, 38(6): 87-92. DOI: 10.19304/j.cnki.issn1000-7180.2021.06.015.
[5]	李倍, 闵丰, 杨军, 等. 一种基于深度学习的目标跟踪加速器[J]. 微电子学与计算机, 2021, 38(8): 53-58. DOI: 10.19304/j.cnki.issn1000-7180.2021.08.009. LI B, MIN F, YANG J, et al. A deep learning object tracking accelerator[J]. Microelectronics & Computer, 2021, 38(8): 53-58. DOI: 10.19304/j.cnki.issn1000-7180.2021.08.009.
[6]	季飞, 王加庆, 刘剑, 等. 适用于硬件高速计算的CNN目标跟踪算法[J]. 微电子学与计算机, 2018, 35(12): 115-118. DOI: 10.19304/j.cnki.issn1000-7180.2018.12.023. JI F, WANG J Q, LIU J, et al. A target tracking algorithm based on CNN for high-speed calculation of hardware[J]. Microelectronics & Computer, 2018, 35(12): 115-118. DOI: 10.19304/j.cnki.issn1000-7180.2018.12.023
[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, 2015, 37(3): 583-596. DOI: 10.1109/TPAMI.2014.2345390.
[8]	WALID W, AWAIS M, AHMED A, et al. Real-time implementation of fast discriminative scale space tracking algorithm[J]. Journal of Real-Time Image Processing, 2021, 18(6): 2347-2360. DOI: 10.1007/s11554-021-01119-6.
[9]	HARE S, SAFFARI A, TORR P H S. Struck: structured output tracking with kernels[C]//Proceedings of 2011 International Conference on Computer Vision. Barcelona: IEEE, 2011: 263-270. DOI: 10.1109/ICCV.2011.6126251.
[10]	KALAL Z, MIKOLAJCZYK K, MATAS J. Tracking-learning-detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 34(7): 1409-1422. DOI: 10.1109/TPAMI.2011.239.
[11]	KRISTAN M, MATAS J, LEONARDIS A, et al. The visual object tracking VOT2015 challenge results[C]//Proceedings of 2015 IEEE International Conference on Computer Vision Workshop (ICCVW). Santiago: IEEE, 2015: 564-586. DOI: 10.1109/ICCVW.2015.79.
[12]	COLLINS R, ZHOU X H, TEH S K. An open source tracking testbed and evaluation web site[C]//Proceedings of IEEE International Workshop on Visual Surveillance and Performance Evaluation of Tracking and Surveillance. IEEE, 2005.