基于特征与区域定位增强的遥感舰船目标检测

doi:10.19678/j.issn.1000-3428.0065425

摘要/Abstract

摘要：

高分辨率遥感图像在海上监视、海上搜救、海上运输等军用和民用领域的舰船检测方面有着广泛的应用。然而高分辨率光学遥感图像舰船目标检测通常存在背景复杂、目标方向任意、尺度多变等问题，导致检测精度不高。提出一种基于特征和区域定位增强的旋转检测算法RetinaNet-MPD。通过添加一个多尺度特征融合模块，充分融合不同尺度、不同层级的特征信息，以增强不同尺度特征图的特征表示能力。针对复杂背景下的舰船目标检测，提出极化双重注意力网络，通过在注意力网络后加入极化函数，充分提取目标的关键特征，同时抑制不相关信息，以有效区分目标和背景。此外，为更准确地定位舰船目标，在对正负样本进行训练时采用一种动态锚学习方法，从而动态选择目标区域内具有良好定位潜力的高质量锚，提高舰船目标检测精度。实验结果表明，RetinaNet-MPD算法在DOTA舰船和HRSC2016数据集上的检测精度分别为89.3%和85.8%，相比现有旋转目标检测算法的检测精度有所提升。

关键词: 高分辨率遥感图像, 舰船目标检测, 多尺度特征融合, 极化双重注意力网络, 动态锚学习

Abstract:

The use of high-resolution remote sensing imagery for ship detection has a wide range of applications in military and civilian fields, such as maritime surveillance, search and rescue, and transportation.However, in high-resolution optical remote sensing images, complex environment as well as arbitrary directions and variable scales of ship targets lead to poor detection accuracy.To address these limitations, a rotation detection algorithm, known as the RetinaNet-MPD, is proposed based on feature and region localization enhancement.First, the RetinaNet-MPD adds a multi-scale feature fusion module, which entirely integrates feature information at different scales and levels, to enhance the feature representation ability of feature maps at different scales.Second, a Polarized Dual-Attention Network(PDANet) module is proposed for ship target detection in a complex environment.By adding a polarization function after the attention network, the key features of the target are entirely extracted, and irrelevant information is suppressed to effectively distinguish the target from its surrounding.In addition, a Dynamic Anchor Learning(DAL) method is adopted when training the positive and negative samples to dynamically select high-quality anchors with good localization potential in the target region and improve the accuracy and precision of ship target detection.The experimental results show that RetinaNet-MPD algorithm achieved detection accuracy of 89.3% and 85.8% on the DOTA-Ship and HRSC2016 data sets, respectively.Consequently, the average detection accuracy was improved effectively compared with other existing rotating-target detection models.

Key words: high-resolution remote sensing imaging, ship target detection, multi-scale feature fusion, Polarized Dual Attention Network(PDANet), Dynamic Anchor Learning(DAL)

宋志娜, 李莎, 杨建明, 徐川. 基于特征与区域定位增强的遥感舰船目标检测[J]. 计算机工程, 2023, 49(8): 257-264.

Zhina SONG, Sha LI, Jianming YANG, Chuan XU. Remote Sensing Ship Target Detection Based on Feature and Region Localization Enhancement[J]. Computer Engineering, 2023, 49(8): 257-264.

http://www.ecice06.com/CN/Y2023/V49/I8/257

图/表 10

图1 两种检测模式的差别

Fig.1 The difference between two detection modes

图2 RetinaNet-MPD网络结构

Fig.2 Structure of RetinaNet-MPD network

图3 多尺度特征融合模块

Fig.3 Multi-scale feature fusion module

图4 双重注意力网络结构

Fig.4 Structure of dual attention network

图5 极化双重注意力网络的结构

Fig.5 Structure of polarized dual attention network

表1 不同算法在HRSC2016数据集上的检测结果对比

Table 1 Comparison of detection results of different algorithms on the HRSC2016 data set %

算法	AP
R2CNN	73.0
RRPN	78.7
RetinaNet-OBB	81.2
ROI Transformer	85.6
R3Det	88.9
RetinaNet-MPD	89.3

表2 不同算法在DOTA-ship数据集上的检测结果对比

Table 2 Comparison of detection results of different algorithms on the DOTA-ship data set %

算法	AP
R2CNN	55.8
RRPN	57.3
RetinaNet-OBB	79.7
ROI Transformer	82.9
R3Det	84.8
RetinaNet-MPD	85.8

图6 不同算法在DOTA-ship数据集上的检测结果

Fig.6 Detection results of different algorithms on the DOTA-ship data set

表3 RetinaNet-MPD算法中不同模块对检测结果的影响

Table 3 Influence of different modules in RetinaNet-MPD algorithm on detection results %

模块与指标	基线方法	添加不同模块
MFF	—	√	—	—	√	√
PDANet	—	—	√	—	√	√
DAL	—	—	—	√	—	√
AP in HRSC2016	81.2	82.7	83.0	87.3	83.9	89.3
AP in DOTA-ship	79.7	80.8	81.1	83.9	82.2	85.8

图7 本文算法与Baseline算法在HRSC2016数据集上的检测结果

Fig.7 Detection results of algorithm in this paper and Baseline algorithm on HRSC2016 data set

参考文献 26

1	HAO Y T. Research on multi-feature and machine learning hierarchical pedestrian detection method based on deep learning. Journal of Physics: Conference Series, 2021, 1748(2): 022001. doi: 10.1088/1742-6596/1748/2/022001
2	DING J, XUE N, LONG Y, et al. Learning RoI transformer for oriented object detection in aerial images[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2020: 2844-2853.
3	黄凤琪, 陈明, 冯国富. 基于可变形卷积的改进YOLO目标检测算法. 计算机工程, 2021, 47(10): 269-275, 282 URL
	HUANG F Q, CHEN M, FENG G F. Improved YOLO object detection algorithm based on deformable convolution. Computer Engineering, 2021, 47(10): 269-275, 282 URL
4	宋志娜, 眭海刚, 李永成. 高分辨率可见光遥感图像舰船目标检测综述. 武汉大学学报(信息科学版), 2021, 46(11): 1703- 1715. URL
	SONG Z N, SUI H G, LI Y C. A survey on ship detection technology in high-resolution optical remote sensing images. Geomatics and Information Science of Wuhan University, 2021, 46(11): 1703- 1715. URL
5	尹雅, 黄海, 张志祥. 基于光学遥感图像的舰船目标检测技术研究. 计算机科学, 2019, 463)(3): 82- 87. URL
	YIN Y, HUANG H, ZHANG Z X. Research on ship detection technology based on optical remote sensing image. Computer Science, 2019, 463)(3): 82- 87. URL
6	刘俊, 姜涛, 徐小康, 等. 基于轻量化深度网络的舰船目标识别技术研究. 无线电工程, 2019, 49(12): 1025- 1030. URL
	LIU J, JIANG T, XU X K, et al. Research on ship target recognition technology based on lightweight deep network. Radio Engineering, 2019, 49(12): 1025- 1030. URL
7	REN S Q, HE K M, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137- 1149. doi: 10.1109/TPAMI.2016.2577031
8	CAI Z W, VASCONCELOS N. Cascade R-CNN: delving into high quality object detection[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2018: 6154-6162.
9	HE K M, GKIOXARI G, DOLLÁR P, et al. Mask R-CNN[C]//Proceedings of IEEE International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2017: 2980-2988.
10	REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2016: 779-788.
11	LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020, 42(2): 318- 327. doi: 10.1109/TPAMI.2018.2858826
12	LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot MultiBox detector[M]//LEIBE B, JIRI MATAS J, SEBE N, et al. Computer vision-ECCV 2016. Berlin, Germany: Springer international publisher, 2016: 21-37.
13	YANG X, YAN J C, FENG Z M, et al. R3Det: refined single-stage detector with feature refinement for rotating object. Proceedings of the AAAI Conference on Artificial Intelligence, 2021, 35(4): 3163- 3171. doi: 10.1609/aaai.v35i4.16426
14	WU Y F, ZHAO W, ZHANG R F, et al. AMR-net: arbitrary-oriented ship detection using attention module, multi-scale feature fusion and rotation pseudo-label. IEEE Access, 2021, 9, 68208- 68222. doi: 10.1109/ACCESS.2021.3075857
15	周旗开, 张伟, 李东锦, 等. 基于改进YOLOv5s的光学遥感图像舰船分类检测方法. 激光与光电子学进展, 2022, 59(16): 466- 473. URL
	ZHOU Q K, ZHANG W, LI D J, et al. Ship classification and detection method for optical remote sensing images based on improved YOLOv5s. Laser & Optoelectronics Progress, 2022, 59(16): 466- 473. URL
16	LIU Z K, WANG H Z, WENG L B, et al. Ship rotated bounding box space for ship extraction from high-resolution optical satellite images with complex backgrounds. IEEE Geoscience and Remote Sensing Letters, 2016, 13(8): 1074- 1078.
17	LIU Z K, HU J G, WENG L B, et al. Rotated region based CNN for ship detection[C]//Proceedings of IEEE International Conference on Image Processing. Washington D. C., USA: IEEE Press, 2018: 900-904.
18	XIA G S, BAI X, DING J, et al. DOTA: a large-scale dataset for object detection in aerial images[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2018: 3974-3983.
19	LIU Z K, YUAN L, WENG L B, et al. A high resolution optical satellite image dataset for ship recognition and some new baselines[C]//Proceedings of the 6th International Conference on Pattern Recognition Applications and Methods. Porto, Portugal: [s. n. ], 2017: 324-331.
20	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2016: 770-778.
21	LIN T Y, DOLLÁR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2017: 936-944.
22	PANG J M, CHEN K, SHI J P, et al. Libra R-CNN: towards balanced learning for object detection[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2020: 821-830.
23	FU J, LIU J, TIAN H J, et al. Dual attention network for scene segmentation[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2020: 3141-3149.
24	MING Q, ZHOU Z Q, MIAO L J, et al. Dynamic anchor learning for arbitrary-oriented object detection. Proceedings of the AAAI Conference on Artificial Intelligence, 2021, 35(3): 2355- 2363.
25	JIANG Y, ZHU X, WANG X, et al. R2CNN: rotational region CNN for orientation robust scene text detection[EB/OL]. [2022-07-01]. https://arxiv.org/abs/1706.09579.
26	MA J Q, SHAO W Y, YE H, et al. Arbitrary-oriented scene text detection via rotation proposals. IEEE Transactions on Multimedia, 2018, 20(11): 3111- 3122.

[1]	任家豪, 张光华, 乔钢柱, 武秀萍. 多尺度特征融合的头影标志点检测[J]. 计算机工程, 2023, 49(3): 271-279.
[2]	赫晓慧, 宋定君, 李盼乐, 田智慧, 周广胜. 融合多尺度特征的遥感影像道路提取方法[J]. 计算机工程, 2022, 48(8): 196-205.
[3]	于敏, 屈丹, 司念文. 改进的RetinaNet目标检测算法[J]. 计算机工程, 2022, 48(8): 249-257.
[4]	白宗文, 弋婷婷, 周美丽, 魏嵬. 基于多尺度特征融合的人脸图像修复方法[J]. 计算机工程, 2021, 47(5): 213-220,228.

选择文件类型/文献管理软件名称

选择包含的内容