为了能够更好地适应水面无人艇复杂的跟踪情况，针对fDDST算法不能处理水面目标被严重遮挡和视野中丢失目标以及帧间目标位置发生突变的问题，对原fDSST算法进行大部分改进。改进水面红外目标搜索区域的方法，将Edge boxes与水天线结合，限定目标检索区域；并结合目标速度和尺度的变化对原算法中滤波器模版的更新率进行改进，同时考虑最大响应值和峰值波动情况对滤波器更新机制进行调整。实验结果证明，改进后的算法能够适应帧间目标大范围的突变和严重遮挡的问题，能保持较好跟踪效果而不跟丢，提高了跟踪算法的鲁棒性，加快了跟踪速度，并且对常见目标跟踪问题仍有很好的跟踪性能。
In order to better adapt to the complex tracking situation of surface unmanned craft, the fDDST algorithm cannot deal with the problems that surface target is seriously blocked, the target is lost in the field of vision and the target position changes between frames, so the original fDSST algorithm is partially improved. In this paper, the method of infrared target search area on water surface is improved, Edge boxes are combined with water antenna to limit the target search area. And the updating rate of the filter template in the original algorithm is improved by combining the target velocity and scale variation, and the updating mechanism of the filter template is adjusted by considering the maximum response value and peak fluctuation. The experimental results show the improved algorithm can adapt to the problems of large range of interframe target mutation and serious occlusion, it can maintain a good tracking effect without losing, improve the robustness of the tracking algorithm, accelerate the tracking speed, and still have a good tracking performance for common target tracking problems.
2022,44(10): 68-72 收稿日期：2021-06-28
DOI：10.3404/j.issn.1672-7649.2022.10.013
分类号：TP399
基金项目：国家自然科学基金资助项目（51709062）
作者简介：管凤旭(1973 - )，男，博士，副教授，研究方向为无人系统自主控制、智能监控与目标跟踪
参考文献：
[1] FEFILATYEV S, GOLDGOF D, SHREVE M, et al. Detection and tracking of ships in open sea with rapidly moving buoy-mounted camera system[J]. Ocean Engineering, 2012, 54(11): 1–12
[2] WU J, MAO S, WANG X, et al. Ship target detection and tracking in cluttered infrared imagery[J]. Optical Engineering, 2011, 50(5): 057207–057207-12
[3] 王悦行, 陈晓冬, 赵克军. 海天背景下红外舰船目标实时跟踪算法研究[J]. 弹箭与制导学报, 2016, 36(1): 165–170
[4] ASHA C S, NARASIMHADHAN A V. Adaptive learning rate for visual tracking using correlation filters[J]. Procedia Computer Science, 2016, 89: 614–622
[5] LIU S, ZHANG T Z, CAO X C, et al. Structural correlation filter for robust visual tracking[C] //Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. LosAlamitos: IEEE Computer Society Press, 2016: 4312–4320.
[6] 段伟伟, 杨学志, 方帅, 等. 分块核化相关滤波目标跟踪[J]. 计算机辅助设计与图形学学报, 2016, 28(7): 1160–1168
[7] 刘威, 赵文杰, 李成. 时空上下文学习长时目标跟踪[J]. 光学学报, 2016, 36(1): 179–186
[8] MARTIN, DANELLJAN, GUSTAV, et al. Discriminative scale space tracking[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(8): 1561–1575
[9] ZITNICK CL, DOLLAR P. Edge boxes: locating object proposals from edges[C]// European Conference on Computer Vision. Springer, Cham, 2014.
[10] WANG M M, LIU Y, HUANG Z Y. Large margin object tracking with circulant feature maps[C]// Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Los Alamitos: IEEE Computer Society Press, 2017: 4800–4808.
[11] DOLLA R P, ZITNICK C L. Structured forests for fast edge detection[C]// International Conference on Computer Vision, 2013: 1841–1848.
[12] 张勋, 马豪伯, 李昀澄. 行均值梯度与直线拟合联合优化的UUV海面红外图像海天线检测[J]. 应用科技, 2018(4): 6–12
[13] MA C, HUANG J B, YANG X K, et al. Adaptive correlation filters with long-term and short-term memory for object tracking[J]. International Journal of Computer Vision, 2018, 126: 771–796
[14] CORTES C, VAPNIK V. Support vector network[J]. Machine Learning, 1995, 20(3): 273–297
[15] 张微, 康宝生. 相关滤波目标跟踪进展综述[J]. 中国图像图形学报, 2017, 22(8): 1017–1033
