利用ELM-AE和迁移表征学习构建的目标跟踪系统

doi:10.3778/j.issn.1673-9418.2012028

计算机科学与探索 ›› 2022, Vol. 16 ›› Issue (7): 1633-1648.DOI: 10.3778/j.issn.1673-9418.2012028

利用ELM-AE和迁移表征学习构建的目标跟踪系统

杨政¹, 邓赵红¹^,⁺(), 罗晓清², 顾鑫², 王士同¹

1.江南大学人工智能与计算机学院,江苏无锡 214122
2.江苏北方湖光光电有限公司,江苏无锡 214035

收稿日期:2020-12-03 修回日期:2021-01-28 出版日期:2022-07-01 发布日期:2021-02-05
作者简介:杨政（1989—）,男,江苏连云港人,硕士研究生,主要研究方向为特征迁移及其在目标跟踪中的应用。
YANG Zheng, born in 1989, M.S. candidate. His research interests include feature transfer and its application in target tracking.
邓赵红（1981—）,男,安徽蒙城人,教授,博士生导师,CCF杰出会员,主要研究方向为不确定性人工智能及其应用。
DENG Zhaohong, born in 1981, professor, Ph.D. supervisor, distinguished member of CCF. His research interests include uncertainty a.pngicial intelligence and its applications.
罗晓清（1980—）,女,江西南昌人,副教授,主要研究方向为图像融合、模式识别、图像处理等。
LUO Xiaoqing, born in 1980, associate professor. Her research interests include image fusion, pattern recognition, image processing, etc.
顾鑫（1979—）,男,江苏张家港人,博士,高级工程师,主要研究方向为模式识别、人工智能、图像处理技术研究与应用。
GU Xin, born in 1979, Ph.D., senior engineer. His research interests include pattern recognition, a.pngicial intelligence, image processing tech-nology and its application.
王士同（1964—）,男,江苏扬州人,教授,博士生导师,主要研究方向为人工智能、模式识别等。
WANG Shitong, born in 1964, professor, Ph.D. supervisor. His research interests include a.pngicial intelligence, pattern recognition, etc.
基金资助:
国家自然科学基金面上项目(61772239);国家自然科学基金面上项目(61772237)

Target Tracking System Constructed by ELM-AE and Transfer Representation Learning

YANG Zheng¹, DENG Zhaohong¹^,⁺(), LUO Xiaoqing², GU Xin², WANG Shitong¹

1. School of A.pngicial Intelligence and Computer Science, Jiangnan University, Wuxi, Jiangsu 214122, China
2. Jiangsu North Huguang Opto-Electronics Co., Ltd., Wuxi, Jiangsu 214035, China

Received:2020-12-03 Revised:2021-01-28 Online:2022-07-01 Published:2021-02-05
Supported by:
the National Natural Science Foundation of China(61772239);the National Natural Science Foundation of China(61772237)

摘要/Abstract

摘要：

在目标跟踪算法中,特征模型对图像特征的快速学习能力和对跟踪过程中目标特征变化的自适应能力一直是目标跟踪算法的主要研究方向之一。特别是对于基于图像块学习的判别式目标跟踪器而言,这两点已然成为影响跟踪器效率和鲁棒性的决定性因素。然而,现有的大多数同类算法在这两个能力上的性能并不能达到令人满意的效果。为了解决这一问题,提出了一种高效且鲁棒的特征模型。该特征模型首先利用基于极限学习机的自编码器（ELM-AE）对目标和背景图像块的复杂图像特征快速地进行随机特征映射,再利用迁移表征学习（TRL）的迁移学习能力提高随机特征空间的自适应性。将该特征模型命名为基于ELM自编码器和迁移表征学习的特征模型（TRL-ELM-AE）。与原复杂图像特征相比,通过该模型可以获得更加紧凑且具有表达能力的共享特征。从而使得分类器可以快速高效地学习和分类。此外,在目标跟踪过程中,目标与背景通常会随着时间不停地变化。虽然TRL的特征迁移能力已经可以很好地适应这一点,但是为了进一步提高跟踪器的鲁棒性,还采用了一种动态更新训练样本的策略。通过对OTB提出的11项目标跟踪挑战场景进行大量实验和分析,证明了所提的目标跟踪器较现有的目标跟踪器具有显著优势。

关键词: 极限学习机（ELM）, 极限学习机自编码器（ELM-AE）, 迁移表征学习（TRL）, 特征自适应, 高斯朴素贝叶斯分类器（GNBC）, 目标跟踪

Abstract:

In the target tracking algorithm, the feature model’s ability to quickly learn image features and the ability to adapt to changes in target features during tracking has always been one of the main research directions of target tracking algorithms. Especially for discriminative target trackers based on image block learning, these two points have become decisive factors affecting the efficiency and robustness of the tracker. However, the performance of most existing similar algorithms on these two abilities cannot achieve satisfactory results. To solve this problem, an efficient and robust feature model is proposed. The feature model first uses extreme learning machine autoencoder (ELM-AE) to quickly perform random feature mapping on complex image features of the target and background image blocks, and then uses the transfer learning ability of transfer representation learning (TRL) to improve the adaptability of random feature space. The feature model is named transfer representation learning with ELM-AE (TRL-ELM-AE). Compared with original complex image features, this model can provide the classifier with more compact and expressive shared features, so that the classifier can learn and classify more quickly and efficiently. In addition, in the target tracking process, the target and background usually change continuously over time. Although the feature migration capability of TRL can already adapt to this, in order to further improve the robustness of the tracker, a strategy of dynamically updating training samples is adopted. Through a large number of experimental and analysis results on the 11 target tracking challenge scenarios proposed by OTB, it is proven that the proposed target tracker has significant advantages over the existing target tracker.

Key words: extreme learning machine (ELM), extreme learning machine autoencoder (ELM-AE), transfer represen-tation learning (TRL), feature adaptation, Gaussian naive Bayes classifier (GNBC), object tracking

中图分类号:

TP181

杨政, 邓赵红, 罗晓清, 顾鑫, 王士同. 利用ELM-AE和迁移表征学习构建的目标跟踪系统[J]. 计算机科学与探索, 2022, 16(7): 1633-1648.

YANG Zheng, DENG Zhaohong, LUO Xiaoqing, GU Xin, WANG Shitong. Target Tracking System Constructed by ELM-AE and Transfer Representation Learning[J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(7): 1633-1648.

图/表 11

图1 TEA-OT的框架图

Fig.1 Framework diagram of TEA-OT

图2 源域与目标域的样本选取

Fig.2 Sample selection of source domain and target domain

图3 TRL-ELM-AE特征提取模型

Fig.3 TRL-ELM-AE feature extraction model

图4 特征差异性对比图

Fig.4 Diagram of feature difference comparison

图5 分类精确度对比图

Fig.5 Comparison diagram of classification accuracy

图6 特征迁移程度的参数分析

Fig.6 Parameter analysis of feature transfer degree

表1 不同隐藏层节点数 L对FPS的影响

Table 1 Influence of different hidden layer nodes L on FPS

隐藏层节点数 $L$	FPS
100	12.61
200	12.21
300	11.92
400	11.74
500	11.45
600	11.22
700	10.79
800	10.55
900	10.24

表1 不同隐藏层节点数 L对FPS的影响

Table 1 Influence of different hidden layer nodes L on FPS

隐藏层节点数 $L$	FPS
100	12.61
200	12.21
300	11.92
400	11.74
500	11.45
600	11.22
700	10.79
800	10.55
900	10.24

图7 ELM-AE隐藏层节点数的参数分析

Fig.7 Parameter analysis of the number of ELM-AE hidden layer nodes

图8 平均像素误差对比图

Fig.8 Average pixel error comparison chart

图9 平均重叠率对比图

Fig.9 Average overlap rate comparison chart

图10 在复杂变化场景中目标跟踪器的适应性展示

Fig.10 Adaptive display of target tracker in complex changing scenes

参考文献 26

[1]	汤一明, 刘玉菲, 黄鸿. 视觉单目标跟踪算法综述[J]. 测控技术, 2020, 39(8): 21-34.
	TANG Y M, LIU Y F, HUANG H. Survey of single-target visual tracking algorithms[J]. Measurement & Control Techn- ology, 2020, 39(8): 21-34.
[2]	李俊彦, 宋焕生, 张朝阳, 等. 基于视频的多目标车辆跟踪及轨迹优化[J]. 计算机工程与应用, 2020, 56(5): 194-199.
	LI J Y, SONG H S, ZHANG Z Y, et al. Multi-object vehicle tracking and trajectory optimization based on video[J]. Computer Engineering and Applications, 2020, 56(5): 194-199.
[3]	BEDDIAR D R, NINI B, SABOKROU M, et al. Vision-based human activity recognition: a survey[J]. Multimedia Tools and Applications, 2020, 79(41): 30509-30555. DOI URL
[4]	YANG L, GEORGESCU B, ZHENG Y, et al. Prediction based collaborative trackers (PCT): a robust and accurate approach toward 3D medical object tracking[J]. IEEE Transactions on Medical Imaging, 2011, 30(11): 1921-1932. DOI URL
[5]	宁纪锋, 吴成柯. 一种基于纹理模型的Mean Shift目标跟踪算法[J]. 模式识别与人工智能, 2007, 20(5): 612-618.
	NING J F, WU C K. A Mean Shift tracking algorithm based on texture model[J]. Pattern Recognition and A.pngicial Inte-lligence, 2007, 20(5): 612-618.
[6]	KIM D S, KWON J. Moving object detection on a vehicle mounted back-up camera[J]. Sensors, 2016, 16(1): 23. DOI URL
[7]	LI X, ZHENG N N. Adaptive target color model updating for visual tracking using particle filter[C]// Proceedings of the 2004 IEEE International Conference on Systems, The Hague, Oct 10-13, 2004. Washington: IEEE Computer Society, 2004: 3105-3109.
[8]	YANG H, SHAO L, ZHENG F, et al. Recent advances and trends in visual tracking: a review[J]. Neurocomputing, 2011, 74(18): 3823-3831. DOI URL
[9]	HAN B, SIM J, ADAM H. BranchOut: regularization for on-line ensemble tracking with convolutional neural networks[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, Jul 21-26, 2017. Washington: IEEE Computer Society, 2017: 521-530.
[10]	LUO W H, LI X, LI W, et al. Robust visual tracking via transfer learning[C]// Proceedings of the 18th IEEE International Conference on Image Processing, Brussels, Sep 11-14, 2011. Piscataway: IEEE, 2011: 485-488.
[11]	GAO J, LING H B, HU W M, et al. Transfer learning based visual tracking with Gaussian processes regression[C]// LNCS 2014:Proceedings of the 13th European Conference on Com-puter Vision, Zurich, Sep 6-12, 2014. Cham: Springer, 2014: 188-203.
[12]	HUANG G, HUANG G B, SONG S, et al. Trends in extreme learning machines: a review[J]. Neural Networks, 2015, 61:32-48.
[13]	LIU H, SUN F, YU Y. Multitask extreme learning machine for visual tracking[J]. Cognitive Computation, 2014, 6(3):391-404. DOI URL
[14]	KASUN L L C, ZHOU H, HUANG G B, et al. Representational learning with ELMs for big data[J]. IEEE Intelligent Systems, 2013, 28(6): 31-34. DOI URL
[15]	PAN S J, TSANG I W, KWOK J T, et al. Domain adaptation via transfer component analysis[J]. IEEE Transactions on Neural Networks, 2011, 22(2): 199-210. DOI URL
[16]	DENG C, HAN Y, ZHAO B. High-performance visual trac-king with extreme learning machine framework[J]. IEEE Transactions on Cybernetics, 2020, 50(6): 2781-2792. DOI URL
[17]	HUANG G B, ZHU Q Y, SIEW C K. Extreme learning machine: a new learning scheme of feedforward neural net-works[C]// Proceedings of the 2004 IEEE International Joint Conference on Neural Networks, Budapest, Jul 25-29, 2004. Piscataway: IEEE, 2005: 985-990
[18]	JOHNSON W B, LINDENSTRAUSS J. Extensions of Lips-chitz maps into a Hilbert space[J]. Contemporary Mathe-matics, 1984, 26(189): 189-206.
[19]	XU P, DENG Z H, WANG J, et al. Transfer representation learning with TSK fuzzy system[J]. IEEE Transactions on Fuzzy Systems, 2021, 29(3): 649-663. DOI URL
[20]	XIE L, DENG Z, XU P, et al. Generalized hidden mapping transductive transfer learning for recognition of epileptic electroencephalogram signals[J]. IEEE Transactions on Cyber-netics, 2019, 49(6): 2200-2214.
[21]	WU Y, LIM J, YANG M H. Online object tracking: a bench-mark[C]// Proceedings of the 2013 IEEE Conference on Com-puter Vision and Pattern Recognition, Portland, Jun 23-28, 2013. Washington: IEEE Computer Society, 2013: 2411-2418.
[22]	JIA X, LU H C, YANG M H. Visual tracking via adaptive structural local sparse appearance model[C]// Proceedings of the 2012 IEEE Conference on Computer Vision and Pattern Recognition, Providence, Jun 16-21, 2012. Washington: IEEE Computer Society, 2012: 1822-1829.
[23]	HENRIQUES J F, CASEIRO R, MARTINS P, et al. Exploi-ting the circulant structure of tracking-by-detection with kernels[C]// LNCS 7575: Proceedings of the 12th European Conference on Computer Vision, Florence, Oct 7-13, 2012. Berlin, Heidelberg: Springer, 2012: 702-715.
[24]	ZHANG K H, ZHANG L, YANG M H. Real-time compressive tracking[C]// LNCS 7574: Proceedings of the 12th European Conference on Computer Vision, Florence, Oct 7-13, 2012. Berlin, Heidelberg: Springer, 2012: 864-877.
[25]	ZHONG W, LU H C, YANG M H. Robust object tracking via sparsity-based collaborative model[C]// Proceedings of the 2012 IEEE Conference on Computer Vision and Pattern Recognition, Providence, Jun 16-21, 2012. Washington: IEEE Computer Society, 2012: 1838-1845.
[26]	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 URL

利用ELM-AE和迁移表征学习构建的目标跟踪系统

Target Tracking System Constructed by ELM-AE and Transfer Representation Learning

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 11

参考文献 26

相关文章 14

编辑推荐

Metrics

[1]	陈洋, 王士同. 多样性正则化极限学习机的集成方法[J]. 计算机科学与探索, 2022, 16(8): 1819-1928.
[2]	刘艺, 李蒙蒙, 郑奇斌, 秦伟, 任小广. 视频目标跟踪算法综述[J]. 计算机科学与探索, 2022, 16(7): 1504-1515.
[3]	李运寰, 闻继伟, 彭力. 高帧率的轻量级孪生网络目标跟踪[J]. 计算机科学与探索, 2022, 16(6): 1405-1416.
[4]	赵运基, 范存良, 张新良. 融合多特征和通道感知的目标跟踪算法[J]. 计算机科学与探索, 2022, 16(6): 1417-1428.
[5]	程世龙, 谢林柏, 彭力. 梯度导向的通道选择目标跟踪算法[J]. 计算机科学与探索, 2022, 16(3): 649-660.
[6]	张毅, 王士同. 在高斯分布下优化仿射变换的极限学习机[J]. 计算机科学与探索, 2021, 15(4): 690-701.
[7]	叶进, 谢紫琪, 肖庆宇, 宋玲, 李晓欢. 数据中心网络中基于ELM的流簇大小推理机制[J]. 计算机科学与探索, 2021, 15(2): 261-269.
[8]	杨金龙，汤玉，张光南. 卷积特征多伯努利视频多目标跟踪算法[J]. 计算机科学与探索, 2019, 13(11): 1945-1957.
[9]	刘芳，黄光伟，路丽霞，王洪娟，王鑫. 自适应模板更新的鲁棒目标跟踪算法[J]. 计算机科学与探索, 2019, 13(1): 83-96.
[10]	张晶，王旭，范洪博. 时空上下文相似性的TLD目标跟踪算法[J]. 计算机科学与探索, 2018, 12(7): 1169-1181.
[11]	孟凡琨，巨永锋，文常保 . 随机网格回归Monte Carlo双UAVs最优目标协调跟踪[J]. 计算机科学与探索, 2017, 11(3): 450-458.
[12]	侯跃恩，李伟光. 结构稀疏表示分类目标跟踪算法[J]. 计算机科学与探索, 2016, 10(7): 1035-1043.
[13]	张继，王洪元. 面向视频目标的快速稀疏编码跟踪算法[J]. 计算机科学与探索, 2012, 6(8): 760-768.
[14]	陈龙，尹彪. 自助银行服务中暴力异常事件检测[J]. 计算机科学与探索, 2012, 6(4): 377-384.