基于形态自适应网络的无人机目标跟踪方法

刘贞报; 马博迪; 高红岗; 院金彪; 江飞鸿; 张军红; 赵闻

doi:10.7527/S1000-6893.2021.24904

航空学报 >

2021 , Vol. 42 >Issue 4: 524904 - 524904

DOI: https://doi.org/10.7527/S1000-6893.2021.24904

电子电气工程与控制

基于形态自适应网络的无人机目标跟踪方法

刘贞报 ,
马博迪 ,
高红岗 ,
院金彪 ,
江飞鸿 ,
张军红 ,
赵闻

展开

1. 西北工业大学民航学院, 西安 710000;
2. 航空工业第一飞机设计研究院飞控系统设计研究所, 西安 710089

收稿日期: 2020-10-20

修回日期: 2020-12-15

网络出版日期: 2021-04-27

基金资助

国家自然科学基金（52072309）；陕西省重点研发计划（2019ZDLGY14-02-01）；深圳市基础研究资助项目（JCYJ20190806152203506）；航空科学基金（ASFC-2018ZC53026）；国家留学基金创新型人才国际合作培养项目（201906290246）

收起

Adaptive morphological network based UAV target tracking algorithm

LIU Zhenbao ,
MA Bodi ,
GAO Honggang ,
YUAN Jinbiao ,
JIANG Feihong ,
ZHANG Junhong ,
ZHAO Wen

Expand

1. School of Civil Aviation, Northwestern Polytechnical University, Xi'an 710000, China;
2. Flight Control System Design Institute, AVIC the First Aircraft Design Institute, Xi'an 710089, China

Received date: 2020-10-20

Revised date: 2020-12-15

Online published: 2021-04-27

Supported by

National Natural Science Foundation of China (52072309); Key Research and Development Program of Shaanxi (2019ZDLGY14-02-01); Shenzhen Fundamental Research Program (JCYJ20190806152203506); Aeronautical Science Foundation of China (ASFC-2018ZC53026); State Scholarship Fund (201906290246)

Fold

摘要

针对无人机影像目标跟踪过程中常出现的目标方向变化、目标遮挡变化、样本多样性不足等问题，提出了一种基于形态自适应网络的无人机航空影像目标跟踪算法。首先使用基于数据驱动的方法对数据集进行扩增，添加了遮挡样本和多旋转角度样本，提高样本多样性；提出的形态自适应网络模型通过旋转不变约束改进深度置信网络，提取强表征能力的深度特征，使得模型能够自动适应目标形态变化，利用深度特征变换算法获取待检测目标的预定位区域，采用基于Q学习算法的搜索机制对目标进行自适应精准定位，使用深度森林分类器提取跟踪目标的类别信息，得到高精度的目标跟踪结果。在多个数据集上进行了对比实验，实验结果表明该算法能够达到较高的跟踪精度，可以适应目标旋转、目标遮挡等形态变化情况，具有较好的准确性和鲁棒性。

关键词： 目标跟踪; 无人机影像; 深度置信网络; Q学习算法; 目标形态变化

本文引用格式

刘贞报 , 马博迪 , 高红岗 , 院金彪 , 江飞鸿 , 张军红 , 赵闻 . 基于形态自适应网络的无人机目标跟踪方法[J]. 航空学报, 2021 , 42(4) : 524904 -524904 . DOI: 10.7527/S1000-6893.2021.24904

Abstract

To solve the problems such as the change of target direction, change of target occlusion and lack of sample diversity in the process of target tracking based on UAV images, this paper proposes a UAV aerial image target tracking algorithm based on the adaptive morphological network. First, the data-driven method is used to expand datasets, and multi rotation angle samples and occlusion samples are added to improve the diversity of samples. The proposed adaptive morphological network improves the deep belief network by rotating invariant constraints to extract deep features with strong representativeness, which enables the model to automatically adapt to the changes of target morphology. The deep feature transformation algorithm is used to obtain the pre-location area of the target to be detected. The target is located adaptively and accurately by the search agent based on the Q-learning algorithm. The category information of the tracking target is extracted by using the deep forest classifier, and the target tracking results with high precision are obtained. Comparative experiments are then carried out on several datasets. The experimental results show that the algorithm can achieve high tracking accuracy, adapt to the change of target angle and occlusion, and has good accuracy and robustness.

Key words： target tracking; UAV image; deep belief network; Q-learning algorithm; target shape change

参考文献

[1] 梁栋, 高赛, 孙涵, 等. 结合核相关滤波器和深度学习的运动相机中无人机目标检测[J]. 航空学报, 2020, 41(9):323733. LIANG D, GAO S, SUN H, et al. UAV detection in motion cameras combining kernelized correlation filters and deep learning[J]. Acta Aeronautica et Astronautica Sinica, 2020, 41(9):323733(in Chinese).
[2] 杨锋平, 马博迪, 王金荣, 等. 基于深度去噪自动编码器的无人机航空影像目标检测[J]. 西北工业大学学报, 2020, 38(6):1345-1351. YANG F P, MA B D, WANG J R, et al. Target detection of UAV aerial image based on rotational invariant depth denoising automatic encoder[J]. Journal of Northwestern Polytechnical University, 2020, 38(6):1345-1351(in Chinese).
[3] 王通, 黄攀峰, 董刚奇. 启发式多无人机协同路网持续监视轨迹规划[J]. 航空学报, 2020, 41(S1):723753. WANG T, HUANG P F, DONG G Q. Cooperation path planning of multi-UAV in road-network continuous monitoring[J]. Acta Aeronautica et Astronautica Sinica, 2020, 41(S1):723753(in Chinese).
[4] 张伟俊, 钟胜, 王建辉. 基于时空显著性建模的空中飞行器跟踪方法[J]. 航空学报, 2020, 41(3):323388. ZHANG W J, ZHONG S, WANG J H. Airborne target tracking based on spatio-temporal saliency modeling[J]. Acta Aeronautica et Astronautica Sinica, 2020, 41(3):323388(in Chinese).
[5] 赵洲, 黄攀峰, 陈路. 一种融合卡尔曼滤波的改进时空上下文跟踪算法[J]. 航空学报, 2017, 38(2):274-284. ZHAO Z, HUANG P F, CHEN L. A tracking algorithm of improved spatiol-temporal context with Kalman filter[J]. Acta Aeronautica et Astronautica Sinica, 2017, 38(2):274-284(in Chinese).
[6] MA B D, LIU Z B, JIANG F H, et al. Vehicle detection in aerial images using rotation-invariant cascaded forest[J]. IEEE Access, 2019, 7:59613.
[7] WANG D, FANG W, CHEN W, et al. Model update strategies about object tracking:A state of the art review[J]. Electronics, 2019, 8:1207.
[8] HENRIQUES F, CASEIRO R, MARTINS P, et al. High-speed tracking with kernelized correlation filters[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2015, 37(3):583.
[9] DANELLJAN M, GUSTAV H, KHAN F S, et al. Accurate scale estimation for robust visual tracking[C]//British Machine Vision Conference, 2014:329.
[10] DANELLJAN M, GUSTAV H, KHAN F S, et al. Learning spatially regularized correlation filters for visual tracking[C]//2015 IEEE International Conference on Computer Vision. Piscataway:IEEE Press, 2015:4310.
[11] BERTINETTO L, VALMADRE J, HENRIQUES J, et al. Fully-Convolutional Siamese networks for object tracking[C]//European Conference on Computer Vision. Berlin:Springer, 2016:42.
[12] LI B, YAN J, WU W, et al. High performance visual tracking with Siamese region proposal network[C]//2018 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2018:8971.
[13] REN S, HE K M, GIRSHICK R, et al. Faster R-CNN:towards real-time object detection with region proposal networks[C]//International Conference on Neural Information Processing System, 2015:91.
[14] GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2014:1.
[15] LEE H, GROSSE R, RANGANATH R, et al. Convolutional deep belief networks for scalable unsupervised learning of hierarchical representations[C]//Proceedings of the 26th Annual International Conference on Machine Learning, 2009:609.
[16] HINTON G E, OSINDERO S, TEH Y W. A fast learning algorithm for deep belief nets[J]. Neural Computation, 2014, 18(7):1527
[17] MNIH V, Kavukcuoglu K, Silver D, et al. Playing Atari with deep reinforcement learning[C]//Conference on Neural Information Processing Systems, 2013:1.
[18] WEI S, ZHANG L, LI Y, et al. Deep descriptor transforming for image co-localization[C]//International Joint Conference on Artificial Intelligence, 2017:3048.
[19] CAICEDO C, LAZEBNIK S. Active object localization with deep reinforcement learning[C]//International Conference on Computer Vision, 2015:2488.
[20] ZHOU Z H, FENG J. Deep forest:Towards an alternative to deep neural networks[C]//International Joint Conference on Artificial Intelligence, 2017:3553.
[21] MUELLER M, SMITH N, GHANEM B. A benchmark and simulator for UAV tracking[C]//European Conference on Computer Vision. Berlin:Springer, 2016:445.
[22] ZHU P F, WEN L Y, DU D W, et al. Vision meets drones:past, present and future[DB/OL]. arXiv preprint:2001.06303, 2020.
[23] BHAT G, JOHNANDER J, DANELLJAN M, et al. Unveiling the power of deep tracking[C]//European Conference on Computer Vision. Berlin:Springer, 2018:172.
[24] VALMADRE J, BERTINETTO L, HENRIQUES F, et al. End-to-End representation learning for correlation filter based tracking[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2017:5000.
[25] DANELLJAN M, BHAT G, KHAN F S, et al. ECO:Efficient convolution operators for tracking[C]//IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE, 2017:6931.
[26] ZHOU X, YAO C, WEN H, et al. EAST:An efficient and accurate scene text detector[C]//IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE, 2017:2642.
[27] ZHAO F, WANG J, WU Y, et al. Adversarial deep tracking[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2019, 29(7):1998.
[28] DANELLJAN M, BHAT G, KHAN F S, et al. ATOM:Accurate tracking by overlap maximization[C]//IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE, 2019:4655.
[29] 任俊丽, 郭浩, 董亚飞, 等. 自适应尺度突变目标跟踪[J]. 中国图象图形学报, 2020, 25(6):1150-1159. REN J L, GUO H, DONG Y F, et al. Adaptive scale sudden change object tracking[J]. Journal of Image and Graphics,2020, 25(6):1150-1159(in Chinese).
[30] JIA F, MARTIN B, MORRIS J. Non-linear principal components analysis for process fault detection[J]. Computers & Chemical Engineering, 1998, 22(S1):851-854.

Options

文章导航

模态框（Modal）标题

摘要

本文引用格式

Abstract

参考文献