基于检测器与定位器融合的自适应校正跟踪算法

doi:10.16182/j.issn1004731x.joss.21-1273

摘要/Abstract

摘要：

针对传统目标跟踪算法在复杂动态场景中因目标发生遮挡、旋转等多种因素而导致的跟踪失败问题，提出了一种基于检测器与定位器融合的自适应校正跟踪算法。定位器通过提取目标的深度特征训练CNN(convolutional neural network)滤波器进行位置估计，CNN滤波器在原CF2（hierarchical convolutional features for visual tracking)算法的3层卷积特征的基础上加入了2层浅层特征，增强了对目标纹理信息的提取。检测器通过提取目标的HOG(histogram of oriented gradient)特征，结合上下文信息计算置信度评分，用当前帧的平均峰值能量和响应最大值分别与历史均值比较，综合判断是否因为遮挡等因素导致跟踪失败，如果跟踪失败，结合检测器进行目标的重定位，否则对目标进行尺度估计。当模型具有高置信度时，更新模型。实验结果表明：算法距离精度和重叠精度均取得不错的效果。

关键词: 卷积特征, 检测器, 跟踪, 自适应校正, 高置信度

Abstract:

In order to avoid tracking failure caused by occlusion, rotation and other factors in complex dynamic scenes, an adaptive correction tracking algorithm based on detector and locator fusion is proposed. The locator trains a convolutional neural network (CNN) filter for location estimation by extracting the deep features of target. The CNN filter adds two layers of shallow features to the three layers of the convolution features of original CF2 algorithm, which enhances the extraction of target texture information. The detector calculates the confidence score by extracting histogram of oriented gradient(HOG) feature of target and combining the context information. The average peak-to-correlation energy (APCE) and maximum response value of current frame are compared separately with the historical average to comprehensively judge whether the tracking fails are due to occlusion and other factors. If the tracking fails, combine the detector to relocate the target, otherwise estimate the scale of the target. Update the model when the model has high confidence. The experimental results show that the distance accuracy and overlap accuracy of the algorithm are good.

Key words: convolution features, detector, tracking, adaptive correction, high confidence

中图分类号:

TP391.41

郭业才, 刘程. 基于检测器与定位器融合的自适应校正跟踪算法[J]. 系统仿真学报, 2023, 35(4): 709-720.

Yecai Guo, Cheng Liu. Adaptive Correction Tracking Algorithm Based on Detector and Locator Fusion[J]. Journal of System Simulation, 2023, 35(4): 709-720.

图/表 10

图1

图2

图3

图4

图5

图6

图7

图8

表1

图9

参考文献 26

1	李琪, 墨瀚林, 王向东, 等. 短道速滑场景下的多目标跟踪与运动仿真[J]. 系统仿真学报, 2021, 33(5): 1039-1050.
	Li Qi, Mo Hanlin, Wang Xiangdong, et al. Multiple Object Tracking and Kinematic Simulation for Short Track Speed Skating[J]. Journal of System Simulation, 2021, 33(5): 1039-1050.
2	侯志强, 陈立琳, 余旺盛, 等. 基于双模板Siamese网络的鲁棒视觉跟踪算法[J]. 电子与信息学报, 2019, 41(9): 2247-2255.
	Hou Zhiqiang, Chen Lilin, Yu Wangsheng, et al. Robust Visual Tracking Algorithm Based on Siamese Network with Dual Templates[J]. Journal of Electronics & Information Technology, 2019, 41(9): 2247-2255.
3	Bertinetto L, Valmadre J, Henriques J F, et al. Fully-Convolutional Siamese Networks for Object Tracking[C]//European Conference On Computer Vision. Cham: Springer, 2016: 850-865.
4	Li B, Wu W, Wang Q, et al. Siamrpn++: Evolution of Siamese Visual Tracking with Very Deep Networks[C]//IEEE/CVF Conference on Computer Vision and Pattern Recognition. Long Beach: IEEE, 2019: 4282-4291.
5	Liu Y, Li R, Cheng Y, et al. Object Tracking Using Spatio-Temporal Networks for Future Prediction Location[C]//European Conference on Computer Vision. Cham: Springer, 2020: 1-17.
6	Bhat G, Danelljan M, Van Gool L, et al. Know Your Surroundings: Exploiting Scene Information for Object Tracking[C]//European Conference on Computer Vision. Cham: Springer, 2020: 205-221.
7	Danelljan M, Bhat G, Shahbaz Khan F, et al. Eco: Efficient Convolution Operators for TRacking[C]//IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 6638-6646.
8	Dai K, Wang D, Lu H, et al. Visual Tracking via Adaptive Spatially-Regularized Correlation Filters[C]//IEEE/CVF Conference on Computer Vision and Pattern Recognition. Long Beach: IEEE, 2019: 4670-4679.
9	Ma C, Huang J B, Yang X K, et al. Hierarchical Convolutional Features for Visual Tracking[C]//IEEE International Conference on Computer Vision. Santiago, Chile: IEEE, 2015: 3074-3082.
10	Kalal Z, Mikolajczyk K, Matas J. Tracking-Learning-Detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence(S0162-8828), 2011, 34(7): 1409-1422.
11	Hong Z, Chen Z, Wang C, et al. Multi-store Tracker (muster): A Cognitive Psychology Inspired Approach to Object Tracking[C]//IEEE Conference on Computer Vision and Pattern Recognition. Boston: IEEE, 2015: 749-758.
12	Ma C, Yang X, Zhang C, et al. Long-term Correlation Tracking[C]//IEEE Conference on Computer Vision and Pattern Recognition. Boston: IEEE, 2015: 5388-5396.
13	Ma C, Huang J B, Yang X, et al. Adaptive Correlation Filters with Long-term and Short-term Memory for Object Tracking[J]. International Journal of Computer Vision(S0920-5691), 2018, 126(8): 771-796.
14	Ma C, Huang J, Yang X, et al. Robust Visual Tracking via Hierarchical Convolutional Features[J]. Proceedings of the IEEE Transactions on Pattern Analysis and Machine Intelligence(S0162-8828), 2018, 41(11): 2709-2723.
15	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(S0162-8828), 2014, 37(3): 583-596.
16	Lu X, Ma C, Ni B, et al. Adaptive Region Proposal with Channel Regularization for Robust Object Tracking[J]. IEEE Transactions on Circuits and Systems for Video Technology(S 10518215), 2019, 31(4): 1268-1282.
17	卫鑫, 武淑红, 王耀力. 基于深度卷积长短期记忆网络的森林火灾烟雾检测模型[J]. 计算机应用, 2019, 39(10): 2883-2887.
	Wei Xin, Wu Shuhong, Wang Yaoli. Forest Fire Smoke Detection Model Based on Deep Convolution Long Short-term Memory Network[J]. Journal of Computer Applications, 2019, 39(10): 2883-2887.
18	徐亮, 张江, 张晶, 等. 基于VGG网络的鲁棒目标跟踪算法[J]. 计算机工程与科学, 2020, 42(8): 1406-1413.
	Xu Liang, Zhang Jiang, Zhang Jing, et al. A Robust Target Tracking Algorithm Based on VGG Network[J]. Computer Engineering & Science, 2020, 42(8): 1406-1413.
19	魏永强, 杨小军. 深度特征的核相关滤波视觉跟踪[J]. 计算机工程与应用, 2020, 56(15): 147-152.
	Wei Yongqiang, Yang Xiaojun. Kernel Correlation Filtering Visual Tracking of Deep Feature[J]. Computer Engineering and Applications, 2020, 56(15): 147-152.
20	Wang M, Liu Y, Huang Z. Large Margin Object Tracking with Circulant Feature Maps[C]//IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 4021-4029.
21	Zitnick C L, Dollar P. Edge Boxes: Locating Object Proposals from Edges[C]//European Conference on Computer Vision. Cham: Springer, 2014: 391-405.
22	Qi Y, Zhang S, Qin L, et al. Hedged Deep Tracking[C]//IEEE Conference On Computer Vision and Pattern Recognition. Las Vegas: IEEE, 2016: 4303-4311.
23	Danelljan M, Häger G, Khan F S, et al. Discriminative Scale Space Tracking[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence(S0162-8828), 2016, 39(8): 1561-1575.
24	Bertinetto L, Valmadre J, Henriques J F, et al. Fully-convolutional Siamese Networks for Object Tracking[C]//European Conference on Computer Vision. Cham: Springer, 2016: 850-865.
25	Li Y, Zhu J. A Scale Adaptive Kernel Correlation Filter Tracker with Feature Integration[C]//European Conference on Computer Vision. Cham: Springer, 2014: 254-265.
26	Danelljan M, Häger G, Khan F, et al. Accurate Scale Estimation for Robust Visual Tracking[C]//British Machine Vision Conference, September 1-5, 2014. Nottingham: Bmva Press, 2014.

算法	BC	DEF	FM	IPR	IV	LR	MB	OCC	OV	OPR	SV
本文	0.883	0.810	0.822	0.884	0.884	0.760	0.817	0.787	0.707	0.841	0.816
HDT	0.844	0.821	0.802	0.844	0.820	0.740	0.794	0.774	0.663	0.805	0.811
CF2	0.843	0.791	0.792	0.854	0.817	0.763	0.797	0.767	0.677	0.807	0.802
siamFC	0.705	0.661	0.722	0.742	0.703	0.749	0.676	0.661	0.664	0.733	0.731
samf	0.689	0.686	0.667	0.721	0.714	0.678	0.677	0.728	0.627	0.738	0.710
fDSST	0.779	0.611	0.690	0.727	0.746	0.586	0.700	0.640	0.578	0.666	0.667
KCF	0.713	0.617	0.620	0.701	0.719	0.505	0.618	0.630	0.501	0.677	0.639
DSST	0.715	0.561	0.569	0.708	0.738	0.568	0.608	0.614	0.480	0.658	0.648

[1]	张思贤, 杨艺, 张猛, 米鹏博. 高效的多特征自适应相关滤波跟踪器[J]. 系统仿真学报, 2022, 34(8): 1864-1873.
[2]	杜宝林, 朱大昌, 盘意华. 机械臂模糊超螺旋二阶滑模轨迹跟踪控制[J]. 系统仿真学报, 2022, 34(6): 1343-1352.
[3]	娄嘉冀, 马静, 李晓威, 赵越, 宋友斌. 微波雷达高精度跟踪环路建模与仿真技术研究[J]. 系统仿真学报, 2022, 34(12): 2619-2628.
[4]	江达, 蔡志勤, 刘忠振, 彭海军, 吴志刚. 基于强化学习的连续型机械臂自适应跟踪控制[J]. 系统仿真学报, 2022, 34(10): 2264-2271.
[5]	齐晓龙, 杨旭光. 离散异构多自主体系统时变输出编队跟踪控制[J]. 系统仿真学报, 2022, 34(1): 36-44.
[6]	路宏广, 赵树恩. 基于鲁棒模型预测的智能汽车轨迹跟踪控制研究[J]. 系统仿真学报, 2022, 34(1): 153-162.
[7]	吴小龙, 夏甫根, 陈静, 徐佳. 基于能量最优的无人驾驶汽车路径跟踪控制[J]. 系统仿真学报, 2022, 34(1): 163-169.
[8]	周维, 刘宇翔, 廖广平, 马鑫. 结合交并比损失的孪生网络目标跟踪算法研究[J]. 系统仿真学报, 2022, 34(09): 1956-1967.
[9]	张瑞民, 陈巧玉. 基于光滑二阶滑模的机械臂轨迹跟踪控制[J]. 系统仿真学报, 2021, 33(6): 1315-1322.
[10]	李琪, 墨瀚林, 王向东, 李华. 短道速滑场景下的多目标跟踪与运动仿真[J]. 系统仿真学报, 2021, 33(5): 1039-1050.
[11]	冷姚, 赵树恩. 智能车辆横向轨迹跟踪的显式模型预测控制方法[J]. 系统仿真学报, 2021, 33(5): 1177-1187.
[12]	范存礼, 戴娟, 刘海涛, 苏中, 朱翠, 徐文婷. 基于神经网络与分数阶滑模的行星进入段轨迹跟踪控制[J]. 系统仿真学报, 2021, 33(11): 2697-2703.
[13]	赵景波, 朱梁鹏, 刘成晔. 四轮转向智能车辆路径跟踪控制策略研究[J]. 系统仿真学报, 2021, 33(10): 2390-2398.
[14]	李琪, 王向东, 李华. 基于双Kinect传感器的三维人体姿态跟踪方法[J]. 系统仿真学报, 2020, 32(8): 1446-1454.
[15]	邓涛, 李鑫. 基于模型预测控制的智能车辆避障跟踪仿真[J]. 系统仿真学报, 2020, 32(8): 1556-1566.