自适应权值卷积特征的鲁棒目标跟踪算法

doi:10.19665/j.issn1001-2400.2019.01.019

摘要/Abstract

摘要：

针对传统基于固定权值卷积特征的深度学习跟踪算法在部分视频跟踪失败的问题,提出一种新颖的基于响应图和熵函数的评估各卷积神经网络层跟踪性能的方法. 该方法能根据评估结果自动调整各层的权值系数;同时引入边界框检测机制,当跟踪响应最大值小于给定阈值时,采用滑动窗口采样一定数量的边界框,并对边界框进行评估,生成初始建议边界框;最后在初始建议边界框的基础上进行相关滤波跟踪,并给出模型更新策略。将文中算法与其他9种算法在OTB-2013视频数据库上进行跟踪仿真,实验结果表明,所提算法具有较高的中心点距离准确率和跟踪成功覆盖率。

关键词: 目标跟踪, 自适应权值, 相关滤波, 目标检测

Abstract:

To solve the tracking failure problem in some videos caused by traditional deep learning tracking algorithms with fixed weight convolutional features, this paper proposes a novel tracking method combing the response map and the entropy function which considers the performance of each layer of convolutional neural networks and automatically adjusts the weight parameters. At the same time, an EdgeBoxes detection scheme is introduced when the maximum value of tracking response is less than a given threshold. A great number of bounding boxes are extracted by a sliding window and are evaluated by the EdgeBoxes detection scheme which generates the original proposal bounding boxes. Finally, the tracking method based on the correlation filter are conducted on the original proposal bounding boxes with the update scheme given. We have tested the proposed algorithm and nine state-of-the-art approaches on OTB-2013 video databases. Experimental results demonstrate that the proposed method has a higher precision and overlap rate.

Key words: object tracking, adaptive weight, correlation filters, object detection

中图分类号:

TP391

王海军,张圣燕. 自适应权值卷积特征的鲁棒目标跟踪算法[J]. 西安电子科技大学学报, 2019, 46(1): 117-123.

WANG Haijun,ZHANG Shengyan. Robust object tracking via adaptive weight convolutional features[J]. Journal of Xidian University, 2019, 46(1): 117-123.

图/表 4

图1

图2

表1

图3

参考文献 17

[1]	胡秀华, 郭雷, 李晖晖 . 一种利用物体性检测的目标跟踪算法[J]. 西安电子科技大学学报, 2017,44(4):86-94, 111. doi: 10.3969/j.issn.1001-2400.2017.04.016
	HU Xiuhua, GUO Lei, LI Huihui . Object Tracking Algorithm Using Objectness Detection[J]. Journal of Xidian University, 2017,44(4):86-94, 111. doi: 10.3969/j.issn.1001-2400.2017.04.016
[2]	王海军, 葛红娟, 张圣燕 . 在线低秩表示的目标跟踪算法[J]. 西安电子科技大学学报, 2016,43(5):98-104. doi: 10.3969/j.issn.1001-2400.2016.05.018
	WANG Haijun, GE Hongjuan, ZHANG Shengyan . Object Tracking via Online Low Rank Representation[J]. Journal of Xidian University, 2016,43(5):98-104. doi: 10.3969/j.issn.1001-2400.2016.05.018
[3]	LI P X, WANG D, WANG L J , et al. Deep Visual Tracking: Review and Experimental Comparison[J]. Pattern Recognition, 2018,76:323-338. doi: 10.1016/j.patcog.2017.11.007
[4]	BOLME D S, BEVERIDGE J S, DRAPER B A, et al. Visual Object Tracking Using Adaptive Correlation Filters [C]//Proceedings of the 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Washington: IEEE Computer Society, 2010: 2544-2550.
[5]	HENRIQUES J, CASEIRO R, MARITINS 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 pmid: 26353263
[6]	LI Y, ZHU J K. A Scale Adaptive Kernel Correlation Filter Tracker with Feature Integration [C]// Lecture Notes in Computer Science: 8926. Heidelberg: Springer Verlag, 2015: 254-265.
[7]	MA C, HUANG J B, YANG X K, et al. Hierarchical Convolutional Features for Visual Tracking [C]//Proceedings of the 2015 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2015: 3074-3082.
[8]	QI Y K, ZHANG S P, QIN L, et al. Hedged Deep Tracking [C]//Proceedings of the 2016 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Washington: IEEE Computer Society, 2016: 4303-4311.
[9]	ZITNICK C L, DOLLA R P. Edge Boxed: Locating Object Proposals from Edges [C]//Lecture Notes in Computer Science: 8693: Heidelberg: Springer Verlag, 2014: 391-405.
[10]	HUANG D F, LUO L, WEN M, et al. Enable Scale and Aspect Ratio Adaptability in Visual Tracking with Detection Proposals [C]//Proceedings of the 2015 26th British Machine Vision Conference. Durham: British Machine Vision Association, 2015: 185.
[11]	ZHANG J M, MA S G, SCLAROFF S. MEEM: Robust Tracking via Multiple Experts using Entropy Minimization [C]// Lecture Notes in Computer Science: 8694 LNCS. Heidelberg: Springer Verlag, 2014: 188-203.
[12]	NING J F, YANG J M, JIANG S J, et al. Object Tracking via Dual Linear Structured SVM and Explicit Feature Map [C] //Proceedings of the 2016 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Washington: IEEE Computer Society, 2016: 4266-4274.
[13]	MATTHIAS M, NEIL S, BERMARD G. Context-Aware Correlation Filter Tracking [C]// Proceedings of the 2017 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Washington: IEEE Computer Society, 2017: 1387-1395.
[14]	HENRIQUES J F, CASEIRO R, MARTINS P, et al. Exploiting the Circulant Structure of Tracking-by-detection with Kernels [C]// Lecture Notes in Computer Science: 7575 LNCS. Heidelberg: Springer Verlag, 2012: 702-715.
[15]	JIA X, LU H, YANG M H . Visual Tracking via Coarse and Fine Structural Local Sparse Appearance Models[J]. IEEE Transactions on Image Processing, 2016,25(10):4555-4564. doi: 10.1109/TIP.2016.2592701 pmid: 27448350
[16]	ZHANG K, YANG M H . Fast Compressive Tracking[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014,36(10):2002-2015. doi: 10.1109/TPAMI.2014.2315808
[17]	WU Y, LIM J, YANG M H. Online Object Tracking: a Benchmark [C]// Proceedings of the 2013 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Washington: IEEE Computer Society, 2013: 2409-2418.

算法	CT	CSK	KCF	DCF_CA	DLSSVM	MEEM	HDT	ASLA	HCFT	Ours
IV	0.295	0.369	0.493	0.530	0.540	0.533	0.557³	0.429	0.560²	0.606¹
OPR	0.297	0.386	0.495	0.536	0.582	0.558	0.584³	0.422	0.587²	0.612¹
SV	0.302	0.350	0.427	0.466	0.494	0.498	0.523³	0.452	0.531²	0.557¹
OC	0.321	0.365	0.514	0.537	0.589	0.552	0.603³	0.376	0.606²	0.609¹
DE	0.345	0.343	0.534	0.576	0.632	0.560	0.627²	0.372	0.626³	0.670¹
MB	0.269	0.305	0.497	0.532	0.578	0.541	0.614²	0.258	0.616¹	0.581³
FM	0.298	0.316	0.459	0.497	0.553	0.553	0.574²	0.247	0.578¹	0.565³
IPR	0.282	0.399	0.497	0.547	0.556	0.535	0.580³	0.425	0.582²	0.605¹
OV	0.359	0.349	0.550	0.543	0.581²	0.606¹	0.569	0.312	0.575³	0.539
BC	0.273	0.421	0.535	0.582	0.592	0.569	0.610³	0.408	0.623²	0.650¹
LR	0.149	0.256	0.272	0.345	0.380	0.401	0.426	0.427³	0.442²	0.496¹