Please wait a minute...
浙江大学学报(工学版)
浙江大学学报(工学版)  2022, Vol. 56 Issue (12): 2379-2391    DOI: 10.3785/j.issn.1008-973X.2022.12.007
计算机技术     
基于通道可靠性和异常抑制的目标跟踪算法
国强1,2(),吴天昊1,2,徐伟1,2,KALIUZHNYMykola1,3
1. 哈尔滨工程大学 信息与通信工程学院，黑龙江 哈尔滨 150001
2. 先进船舶通信与信息技术工业和信息化部重点实验室，黑龙江 哈尔滨 150001
3. 哈尔科夫国立无线电电子大学，乌克兰 哈尔科夫 61166
Target tracking algorithm based on channel reliability and aberrance repression
Qiang GUO1,2(),Tian-hao WU1,2,Wei XU1,2,Mykola KALIUZHNY1,3
1. College of Information and Communication Engineering, Harbin Engineering University, Harbin 150001, China
2. Key Laboratory of Advanced Marine Communication and Information Technology, Ministry of Industry and Information Technology, Harbin 150001, China
3. Kharkiv National University of Radio Electronics, Kharkiv 61166, Ukraine
 全文: PDF(3036 KB)   HTML
摘要：

针对时空正则目标跟踪算法无法有效利用特征，为了缓解边界效应扩大搜索区域导致的滤波器倾向于从背景中学习的问题，提出基于通道可靠性和异常抑制的目标跟踪算法. 构造通道正则项,在训练阶段求解不同特征通道对应的权重，实现对不同特征通道的加权,降低通道冗余并提高定位精度.在目标函数中加入异常抑制正则项，约束当前帧的响应图,实现滤波器模型的平滑约束. 利用交替方向乘子法将求解目标问题转化为求滤波器、辅助因子以及通道权重的最优解. 将所提算法在OTB2015、TempleColor128以及UAV20L公开数据集测试并与其他跟踪算法进行对比. 实验结果表明，所提算法在快速运动、光照变化场景中的跟踪效果稳定，基本满足实时性要求.
关键词： 目标跟踪;  时空正则;  通道正则;  异常抑制;  交替方向乘子法    
Abstract:

A target tracking algorithm based on channel reliability and aberrance repression was proposed, aiming at the problem that the spatial-temporal regularized target tracking algorithm could not effectively use features and the filter was inclined to learn from the background since the algorithm expanded the search area in order to alleviate the boundary effect. The channel regularization term was constructed, and the corresponding weights of different feature channels were solved in the training stage to realize the weighting of different feature channels, which reduced channel redundancy and improved the positioning accuracy. The aberrance repression regularization term was added to the objective function to constrain the response map of the current frame and realize the smooth constraint of the filter model. The alternating direction multiplier method was used to transform the objective problem into the optimal solution of the filter, auxiliary factor and channel weight. The proposed method was tested on public datasets such as OTB2015, TempleColor128 and UAV20L, and compared with other tracking algorithms. Experimental results show that the tracking effect of the proposed algorithm is stable under fast motion as well as illumination variation, and it basically meets the real-time requirements．
Key words: target tracking    spatial-temporal regularization    channel regularization    aberrance repression    alternating direction multiplier method
收稿日期: 2021-09-25 出版日期: 2023-01-03
CLC:  TP 391.4  
基金资助: 国家重点研发计划资助项目（2018YFE0206500）；国家自然科学基金资助项目（62071140）；国家国际科技合作专项资助项目（2015DFR10220）
作者简介: 国强（1972—），男，教授，从事通信对抗研究. orcid.org/0000-0002-8366-7163. E-mail： guoqiang@hrbeu.edu.cn
服务　　
把本文推荐给朋友
加入引用管理器
E-mail Alert
作者相关文章　　
国强
吴天昊
徐伟
KALIUZHNYMykola

引用本文:

国强,吴天昊,徐伟,KALIUZHNYMykola. 基于通道可靠性和异常抑制的目标跟踪算法[J]. 浙江大学学报(工学版), 2022, 56(12): 2379-2391.

Qiang GUO,Tian-hao WU,Wei XU,Mykola KALIUZHNY. Target tracking algorithm based on channel reliability and aberrance repression. Journal of ZheJiang University (Engineering Science), 2022, 56(12): 2379-2391.

链接本文:

https://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2022.12.007    或    https://www.zjujournals.com/eng/CN/Y2022/V56/I12/2379

图 1  不同特征通道的权重可视化
图 2  视频序列异常分析示意图
图 3  基于通道可靠性和异常抑制的目标跟踪算法流程图
图 4  不同算法在OTB2015数据集上的距离精度曲线和重叠成功率曲线下面积曲线
算法 $v $
OTB2015 TempleColor128 UAV20L
帧/s
本研究 21.90 19.48 17.25
BACF 35.44 36.75 27.13
STRCF 24.28 21.31 19.17
SRDCF 7.60 8.42 6.04
SRDCFDecon 2.01 3.09 3.38
ARCF 17.80 16.49 16.08
AutoTrack 28.44 27.52 23.46
DRCF 25.25 29.61 27.12
ECO_HC 59.39 58.64 63.21
HCF 1.97 1.92 6.70
HDT 3.69 3.57 2.88
FAST 72.87 44.94 82.12
SITUP 27.79 19.52 10.61
LCT 22.79 25.26 29.54
KCF 234.19 240.64 235.62
表 1  各算法在3个数据集上的平均跟踪速度对比
图 5  不同算法在TempleColor128数据集上的距离精度曲线和重叠成功率曲线下面积曲线
图 6  不同算法在UAV20L数据集上的距离精度曲线和重叠成功率曲线下面积曲线
算法 DP
IV SV OC DEF MB FM IPR OPR OV BC LR
%
本研究 84.6 85.0 82.6 84.6 84.3 81.7 82.4 86.1 79.9 88.2 74.5
BACF 80.3 76.9 73.0 76.4 74.5 79.0 79.2 78.1 75.6 80.5 74.1
STRCF 83.7 84.0 81.0 84.1 82.6 80.2 81.1 85.0 76.6 87.2 73.7
DeepSRDCF 78.6 81.7 82.2 77.9 82.3 81.4 81.8 83.5 78.1 84.1 70.8
SRDCF 78.1 74.3 72.7 73.0 76.7 76.9 74.2 74.0 60.3 77.5 66.3
SRDCFDecon 83.3 80.3 76.5 75.0 81.4 77.5 77.6 79.7 64.1 85.0 64.4
ARCF 76.3 77.0 73.7 76.7 75.7 76.8 78.5 76.9 67.1 76.0 74.9
AutoTrack 78.3 74.2 73.5 73.5 73.5 74.6 77.7 76.6 69.6 75.5 77.3
DRCF 71.8 67.6 67.0 72.0 71.8 74.5 69.4 69.2 61.1 76.2 62.1
ECO_HC 77.5 79.2 77.7 79.3 77.0 79.9 76.2 80.1 76.4 80.7 84.7
HCF 83.0 79.8 77.6 79.0 80.4 81.5 86.4 81.6 67.7 84.3 83.1
HDT 80.9 77.4 74.4 80.2 78.3 77.9 79.9 78.7 61.6 78.9 84.9
SiamFC 74.1 73.8 72.6 69.3 70.5 74.3 74.2 75.6 66.9 69.0 84.7
FAST 76.7 70.8 70.8 70.0 61.3 64.2 73.0 77.0 61.3 77.3 71.9
SITUP 73.5 74.5 73.8 69.7 70.7 70.2 73.9 76.6 66.9 77.6 68.4
LCT 74.3 67.8 67.8 68.5 67.0 68.1 78.1 74.6 59.2 73.4 53.7
KCF 72.4 63.5 63.2 61.9 60.0 62.1 70.1 67.6 50.0 71.3 56.0
表 2  在不同属性的OTB2015数据集上不同算法的距离精度
算法 AUC
IV SV OC DEF MB FM IPR OPR OV BC LR
%
本研究 66.6 63.6 62.8 61.8 66.4 63.5 60.7 63.7 60.2 66.3 54.6
BACF 62.2 57.2 56.5 57.1 57.5 59.9 58.2 57.8 54.8 60.5 53.2
STRCF 65.2 63.1 61.4 60.5 65.2 62.8 60.2 62.6 58.3 64.7 53.8
DeepSRDCF 62.4 60.7 60.3 56.7 64.2 62.8 58.9 60.7 55.3 62.7 47.5
SRDCF 60.7 55.9 55.4 54.1 59.4 59.7 54.1 54.7 46.1 58.2 49.5
SRDCFDecon 64.7 60.7 58.8 55.2 63.9 60.6 57.3 59.1 51.0 64.1 49.2
ARCF 60.0 56.1 56.0 58.3 60.5 59.1 56.2 55.9 50.0 58.8 51.2
AutoTrack 60.4 54.2 55.5 55.9 58.5 58.3 55.4 55.5 53.4 56.2 54.0
DRCF 56.8 51.7 52.7 54.0 59.4 59.0 52.4 52.8 49.0 57.1 45.2
ECO_HC 60.3 59.2 58.7 58.7 60.4 61.8 55.3 58.7 56.0 60.3 58.9
HCF 55.0 48.5 53.3 53.0 58.5 57.0 56.6 54.0 47.4 58.5 43.9
HDT 52.9 47.7 52.2 54.0 57.7 55.5 53.6 52.8 45.3 54.4 46.3
SiamFC 57.4 55.6 54.7 51.0 55.0 56.8 55.7 55.8 50.6 52.3 59.2
FAST 59.8 52.1 53.6 51.1 50.1 50.7 53.5 56.4 47.8 58.9 50.8
SITUP 55.4 53.1 55.6 50.4 57.4 54.2 54.0 55.8 52.0 57.2 43.4
LCT 51.7 42.8 47.6 48.1 51.6 50.7 52.9 50.5 44.6 52.8 29.9
KCF 48.2 39.5 44.5 43.8 45.9 45.9 46.9 45.3 39.3 49.8 30.7
表 3  在不同属性的OTB2015数据集上不同算法的重叠成功率曲线下面积
图 7  不同算法在不同视频序列上的跟踪结果对比
$ \lambda $ DP/% $ \lambda $ DP/% $ \lambda $ DP/%
0.01 84.9 0.04 86.2 0.07 86.1
0.02 85.9 0.05 86.9 0.08 86.2
0.03 86.3 0.06 85.9 0.09 85.9
表 4  不同通道可靠性惩罚系数下的距离精度
ρ DP/% ρ DP/% ρ DP/%
0.010 86.5 0.060 86.4 0.069 86.6
0.020 86.3 0.065 86.7 0.070 87.0
0.030 85.7 0.066 86.6 0.080 87.1
0.040 86.2 0.067 86.6 0.090 86.5
0.050 87.1 0.068 87.3 — —
表 5  不同异常抑制惩罚系数下的距离精度
算法 DP AUC
%
本研究 87.3 66.1
Baseline+ARR 87.0 65.9
Baseline+CRR 86.9 65.7
Baseline 86.5 65.4
表 6  OTB100数据集上的消融实验结果
1 黄月平, 李小锋, 杨小冈, 等 基于相关滤波的视觉目标跟踪算法新进展[J]. 系统工程与电子技术, 2021, 43 (8): 2051- 2065
HUANG Yue-ping, LI Xiao-feng, YANG Xiao-gang, et al Advances in visual object tracking algorithm based on correlation filter[J]. Systems Engineering and Electronics, 2021, 43 (8): 2051- 2065
doi: 10.12305/j.issn.1001-506X.2021.08.05
2 BOLME D S, BEVERIDGE J R, DRAPER B A, et al. Visual object tracking using adaptive correlation filters [C]// 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. San Francisco: IEEE, 2010: 2544-2550.
3 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
4 DANELLJAN M, KHAN F S, FELSBERG M, et al. Adaptive color attributes for real-time visual tracking [C]// 2014 IEEE Conference on Computer Vision and Pattern Recognition. Columbus: IEEE, 2014: 1090-1097.
5 WANG N Y, SHI J P, YEUNG D Y, et al. Understanding and diagnosing visual tracking systems [C]// 2015 IEEE International Conference on Computer Vision. Santiago: IEEE, 2015: 3101-3109.
6 汤张泳, 吴小俊, 朱学峰 多空间分辨率自适应特征融合的相关滤波目标跟踪算法[J]. 模式识别与人工智能, 2020, 33 (1): 66- 74
TANG Zhang-yong, WU Xiao-jun, ZHU Xue-feng Object tracking with multi-spatial resolutions and adaptive feature fusion based on correlation filters[J]. Pattern Recognition and Artifical Intelligence, 2020, 33 (1): 66- 74
doi: 10.16451/j.cnki.issn1003-6059.202001008
7 熊昌镇, 王润玲, 邹建成 基于多高斯相关滤波的实时跟踪算法[J]. 浙江大学学报:工学版, 2019, 53 (8): 1488- 1495
XIONG Chang-zhen, WANG Run-ling, ZOU Jian-cheng Real-time tracking algorithm based on multiple Gaussian-distirbution correlation filters[J]. Journal of Zhejiang University: Engineering Science, 2019, 53 (8): 1488- 1495
8 GALOOGAHI H K, FAGG A, LUCEY S. Learning backg- round-aware correlation filters for visual tracking [C]// 2017 IEEE International Conference on Computer Vision. Venice: IEEE, 2017: 1144-1152.
9 DANELLJAN M, HÄGER G, KHAN F S, et al. Learning spatially regularized correlation filters for visual tracking [C]// 2015 IEEE International Conference on Computer Vision. Santiago: IEEE, 2015: 4310-4318.
10 LI F, TIAN C, ZUO W M, et al. Learning spatial-temporal regularized correlation filters for visual tracking [C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City: IEEE, 2018: 4904-4913.
11 尹明锋, 薄煜明, 朱建良, 等. 基于通道可靠性的多尺度背景感知相关滤波跟踪算法[J]. 光学学报. 2019, 39(5): 0515002.
YIN Ming-feng, BO Yu-ming, ZHU Jian-liang, et al. Multi-scale context-aware correlation filter tracking algorithm based on channel reliability [J]. Acta Optica Sinica, 2019, 39(5): 0515002.
12 LU X K, MA C, NI B B, et al Adaptive region proposal with channel regularization for robust object tracking[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2021, 31 (4): 1268- 1282
doi: 10.1109/TCSVT.2019.2944654
13 XU T Y , FENG Z H, WU X J, et al. Joint group feature selection and discriminative filter learning for robust visual obj-ect tracking [C]// 2019 IEEE/CVF International Conference on Computer Vision. Seoul: IEEE, 2019: 7950-7960.
14 YE J J, FU C H, LIN F L, et al Multi-regularized correlation filter for UAV tracking and self-localization[J]. IEEE Transactions on Industrial Electronics, 2021, 6004- 6014
15 WANG M M, LIU Y, HUANG Z Y. Large margin object tracking with circulant feature maps [C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 4800-4808.
16 HUANG Z Y, FU C H, LI Y M, et al. Learning aberrance repressed correlation filters for real-time UAV tracking [C]// Proceedings of the 2019 IEEE/CVF International Conference on Computer Vision. Seoul: IEEE, 2019: 2891-2900.
17 LI Y M, FU C H, DING F Q, et al. AutoTrack: towards high-performance visual tracking for UAV with automatic spatio-temporal regularization [C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle: IEEE, 2020: 11920-11929.
18 姜文涛, 刘晓璇, 涂潮, 等 空间异常适应性的目标跟踪[J]. 模式识别与人工智能, 2021, 34 (5): 473- 484
JIANG Wen-tao, LIU Xiao-xuan, TU Chao, et al Spatially abnormal adaptive target tracking[J]. Pattern Recognition and Artificial Intelligence, 2021, 34 (5): 473- 484
doi: 10.16451/j.cnki.issn1003-6059.202105010
19 BOYD S, PARIKH N, CHU E, et al Distributed optimization and statistical learning via the alternating direction method of multipliers[J]. Foundations and Trends in Machine Leaning, 2010, 3 (1): 1- 122
doi: 10.1561/2200000016
20 WU Y, LIM J, YANG M H Object tracking benchmark[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37 (9): 1834- 1848
doi: 10.1109/TPAMI.2014.2388226
21 LIANG P P, BLASCH E, LING H B. Encoding color info- rmation for visual tracking: algorithms and benchmark [J]. IEEE Transactions on Image Processing, 2015, 24(12): 5630-5644.
22 MUELLER M, SMITH N, GHANEM B. A benchmark and simulator for UAV tracking [C]// 2016 European Conference on Computer Vision. Amsterdam: Springer, 2016: 445–461.
23 MA C, HUANG J B, YANG X, et al. Hierarchical convolutional features for visual tracking [C]// 2015 IEEE International Conference on Computer Vision. Santiago: IEEE, 2015: 3074-3082.
24 QI Y, ZHANG S, QIN L, et al. Hedged deep tracking [C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas: IEEE, 2016: 4303-4311.
25 DANELLJAN M, HÄGER G, KHAN F S, et al. Convolutional features for correlation filter based visual tracking [C]// 2015 IEEE International Conference on Computer Vision Workshop. Santiago: IEEE, 2015: 621-629.
26 DANELLJAN M, BHAT G, KHAN F S, et al. ECO: efficient convolution operators for tracking [C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 6931-6939.
27 DANELLJAN M, HÄGER G, KHAN F S, et al. Adaptive decontamination of the training set: a unified formulation for discriminative visual tracking [C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas : IEEE, 2016: 1430-1438.
28 MA C, YANG X K, ZAHNG C Y, et al. Long-term correlation tracking [C]// 2015 IEEE Conference on Computer Vision and Pattern Recognition. Boston: IEEE, 2015: 5388-5396.
29 MA H Y, ACTON S T, LIN Z L SITUP: scale invariant tracking using average peak-to-correlation energy[J]. IEEE Transactions on Image Processing, 2020, 29: 3546- 3557
doi: 10.1109/TIP.2019.2962694
30 MA H Y, LIN Z L, ACTON S T FAST: fast and accurate scale estimation for tracking[J]. IEEE Signal Processing Letters, 2020, 27: 161- 165
doi: 10.1109/LSP.2019.2963147
31 FU C H, XU J T, LIN F L, et al Object saliency-aware dual regularized correlation filter for real-time aerial tracking[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, 58 (12): 8940- 8951
doi: 10.1109/TGRS.2020.2992301
[1] 王海军,张圣燕,杜玉杰. 响应和滤波器偏差感知约束的无人机目标跟踪算法[J]. 浙江大学学报(工学版), 2022, 56(9): 1824-1832.
[2] 华夏,王新晴,芮挺,邵发明,王东. 视觉感知的无人机端到端目标跟踪控制技术[J]. 浙江大学学报(工学版), 2022, 56(7): 1464-1472.
[3] 陈宏鑫,张北,王春香,杨明. 基于自适应随动机构的机器人目标跟随[J]. 浙江大学学报(工学版), 2022, 56(6): 1071-1078.
[4] 张迅,李建胜,欧阳文,陈润泽,汲振,郑凯. 融合运动信息和跟踪评价的高效卷积算子[J]. 浙江大学学报(工学版), 2022, 56(6): 1135-1143, 1167.
[5] 陈小波,陈玲,梁书荣,胡煜. 重尾非高斯定位噪声下鲁棒协同目标跟踪[J]. 浙江大学学报(工学版), 2022, 56(5): 967-976.
[6] 金立生,华强,郭柏苍,谢宪毅,闫福刚,武波涛. 基于优化DeepSort的前方车辆多目标跟踪[J]. 浙江大学学报(工学版), 2021, 55(6): 1056-1064.
[7] 周金海,周世镒,常阳,吴耿俊,王依川. 基于超宽带雷达基带信号的多人目标跟踪[J]. 浙江大学学报(工学版), 2021, 55(6): 1208-1214.
[8] 宋鹏,杨德东,李畅,郭畅. 整体特征通道识别的自适应孪生网络跟踪算法[J]. 浙江大学学报(工学版), 2021, 55(5): 966-975.
[9] 王康豪,殷海兵,黄晓峰. 基于策略梯度的目标跟踪方法[J]. 浙江大学学报(工学版), 2020, 54(10): 1923-1928.
[10] 荆丹翔,韩军,徐志伟,陈鹰. 基于成像声呐的水下多目标跟踪研究[J]. 浙江大学学报(工学版), 2019, 53(4): 753-760.
[11] 王海军, 葛红娟, 张圣燕. 基于核协同表示的快速目标跟踪算法[J]. 浙江大学学报(工学版), 2017, 51(2): 399-407.
[12] 卢维, 项志宇, 于海滨, 刘济林. 基于自适应多特征表观模型的目标压缩跟踪[J]. 浙江大学学报(工学版), 2014, 48(12): 2132-2138.
[13] 王安定, 裘渔洋, 王秀萍, 余燕平, 李式巨. 基于Toeplitz降维子矩阵的空间多目标跟踪算法[J]. J4, 2012, 46(4): 739-743.
[14] 戴渊明, 韦巍, 林亦宁. 基于颜色纹理特征的均值漂移目标跟踪算法[J]. J4, 2012, 46(2): 212-217.
[15] 许雪梅, 李丽娴, 张键洋, 倪兰, 黄征宇, 曹建. 透明液体药剂中可见异物跟踪算法[J]. J4, 2012, 46(10): 1822-1830.