基于语义分割实现的SAR图像舰船目标检测

doi:10.12305/j.issn.1001-506X.2022.04.16

摘要/Abstract

摘要：

基于深度学习实现的目标检测方法在自然图像中取得非常大的成功, 而将诸多方法运用于合成孔径雷达(synthetic aperture radar, SAR)图像舰船目标检测逐步成为新的趋势。如何将已有方法改进并与SAR图像的特点相结合完成特定的检测任务, 已经成为当前主要的研究方向。不同于当前已有方法, 本文对存在的深度学习SAR图像舰船目标检测方法进行了再思考, 提出了基于语义分割实现的检测、分割一体化方法。通过语义分割实现的检测方式能够有效地避免当前诸多检测网络的复杂解码过程, 具有生成的预测框更加贴合目标、精度以及召回率更高等特点。该方法虽属于无锚框检测, 但实验结果表明, 达到了双阶段检测效果, 且具有更加精细化的分割结果, 适用于复杂背景检测与分割问题。

关键词: 合成孔径雷达, 语义分割, 舰船检测, 卷积神经网络, 深度学习

Abstract:

Object detection methods based on deep learning have achieved outstanding success in natural images, and the application of many methods to ship detection in synthetic aperture radar (SAR) images has gradually become a new trend. How to improve the existing methods and combine them with the characteristics of SAR images to complete specific detection tasks has become the main research direction at present. Different from the current detection methods, this paper rethinks the existing deep learning-based SAR image ship detection algorithms and proposes an integrated detection and segmentation method based on semantic segmentation. This detection method realized by semantic segmentation can effectively avoid the complicated decoding process of many detection networks. Besides, it has the characteristics of predicted bounding boxes generated in the prediction stage that are more suitable for targets, and has a higher precision and recall rate compared with other methods. Although this method belongs to the anchor-free detection field, the experimental results show it achieves a two-stage detection effect, and has more refined segmentation results, which is suitable for complicated background noise-based detection and segmentation.

Key words: synthetic aperture radar (SAR), semantic segmentation, ship detection, convolutional neural network (CNN), deep learning

中图分类号:

TP751

陈冬, 句彦伟. 基于语义分割实现的SAR图像舰船目标检测[J]. 系统工程与电子技术, 2022, 44(4): 1195-1201.

Dong CHEN, Yanwei JU. Ship object detection SAR images based on semantic segmentation[J]. Systems Engineering and Electronics, 2022, 44(4): 1195-1201.

图/表 8

图1

图2

表1

表2

表3

图3

图4

图5

参考文献 41

1	VILLANO M , KRIEGER G , STEINBRECHER U , et al. Simultaneous single-/dual- and quad-pol SAR imaging over swaths of different widths[J]. IEEE Trans. on Geoscience and Remote Sensing, 2020, 25 (3): 2096- 2103.
2	焦李成, 张向荣, 侯彪, 等. 智能SAR图像处理与解译[M]. 北京: 科学出版社, 2008.
	JIAO L C , ZHANG X R , HOU B , et al. Intelligent SAR image processing and interpretation[M]. Beijing: China Science Publishing & Media Ltd., 2008.
3	SHEN H F , ZHOU C X , LI J , et al. SAR image despeckling employing a recursive deep CNN prior[J]. IEEE Trans. on Geoscience and Remote Sensing, 2021, 59 (1): 273- 286. doi: 10.1109/TGRS.2020.2993319
4	WANG P Y , ZHANG H , PATEL V M . SAR image despeckling using a convolutional neural network[J]. IEEE Signal Processing Letters, 2017, 24 (12): 1763- 1767. doi: 10.1109/LSP.2017.2758203
5	CUI Z Y , WANG X Y , LIU N Y , et al. ship detection in large-scale SAR images via spatial shuffle-group enhance attention[J]. IEEE Trans. on Geoscience and Remote Sensing, 2021, 59 (1): 379- 391. doi: 10.1109/TGRS.2020.2997200
6	LIU L, CHEN G W, PAN Z X, et al. Inshore ship detection in SAR images based on deep neural networks[C]//Proc. of the IEEE International Geoscience and Remote Sensing Symposium, 2018: 5-28.
7	陈冬, 句彦伟. 基于改进型YOLOv3的SAR图像舰船目标检测[J]. 系统工程与电子技术, 2021, 43 (4): 937- 943.
	CHEN D , JU Y W . SAR ship detection based on improved YOLOv3[J]. Systems Engineering and Electronics, 2021, 43 (4): 937- 943.
8	LI W K, ZOU B, XIN Y, et al. An improved CFAR scheme for man-made target detection in high resolution SAR images[C]//Proc. of the IEEE International Geoscience and Remote Sensing Symposium, 2018: 2829-2832.
9	杨国铮, 禹晶, 肖创柏, 等. 基于形态字典学习的复杂背景SAR图像舰船尾迹检测[J]. 自动化学报, 2017, 43 (10): 1713- 1725.
	YANG G Z , YU J , XIAO C B , et al. Ship wake detection in SAR images with complex background using morphological dictionary learning[J]. Acta Automatic Sinica, 2017, 43 (10): 1713- 1725.
10	ZHU J W , QIU X L , PAN Z X , et al. Projection shape template-based ship target recognition in TerraSAR-X images[J]. IEEE Geoscience and Remote Sensing Letters, 2017, 14 (2): 222- 226. doi: 10.1109/LGRS.2016.2635699
11	CHEN G , LI G , LIU Y , et al. SAR image despeckling based on combination of fractional-order total variation and nonlocal low rank regularization[J]. IEEE Trans. on Geoscience and Remote Sensing, 2020, 58 (3): 2056- 2070. doi: 10.1109/TGRS.2019.2952662
12	MOLINI A B, VALSESIA D, FRACASTORO G, et al. Towards deep unsupervised SAR despeckling with blind-spot convolutional neural networks[C]//Proc. of the IEEE International Geoscience and Remote Sensing Symposium, 2020: 2507-2510.
13	DALALSASSO E , DENIS L , TUPIN F . SAR2SAR: a semi-supervised despeckling algorithm for SAR images[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2021, 14, 4321- 4329. doi: 10.1109/JSTARS.2021.3071864
14	GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proc. of the Computer Vision and Pattern Recognition, 2014: 580-587.
15	GIRSHICK R. Fast R-CNN[C]//Proc. of the International Conference on Computer Vision, 2015: 1440-1448.
16	REN S Q , HE K M , GIRSHICK R , et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Trans. on Pattern Analysis & Machine Intelligence, 2017, 39 (6): 1137- 1149.
17	CAI Z, VASCONCELOS N. Cascade R-CNN: delving into high quality object detection[C]//Proc. of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018: 6154-6162.
18	HE K , GKIOXARI G , DOLLÁR P , et al. Mask R-CNN[J]. IEEE Trans. on Pattern Analysis and Machine Intelligence, 2020, 42 (2): 386- 397. doi: 10.1109/TPAMI.2018.2844175
19	LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot multibox detector[C]//Proc. of the European Conference on Computer Vision, 2016: 21-37.
20	LI Z X , ZHOU F Q . FSSD: feature fusion single shot multibox detector[J]. Computer Vision and Pattern Recognition, 2018, 36 (7): 356- 366.
21	REDMON J, DIVVALA S K, GIRSHICK R, et al. You only look once: unified, real-time object detection[C]//Proc. of the Computer Vision and Pattern Recognition, 2016: 779-788.
22	REDMON J, FARHADI A. YOLO9000: better, faster, stronger[C]//Proc. of the Computer Vision and Pattern Recognition, 2017: 6517-6525.
23	REDMON J, FARHADI A. YOLOv3: an incremental improvement[EB/OL]. [2021-5-10]. https://arxiv.org/abs/1804.02767v1.
24	BOCHKOVSKIY A, WANG C, LIAO H. YOLOv4: optimal speed and accuracy of object detection[EB/OL]. [2021-5-10]. https://arxiv.org/abs/2004.10934v1.
25	LI R , WANG X D , WANG J , et al. SAR target recognition based on efficient fully convolutional attention block CNN[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19, 4005905.
26	WU S N, WANG K, OUYANG Y W. Study on small samples SAR image recognition detection method based on transfer CNN[C]//Proc. of the 3rd International Conference on Electronic Information Technology and Computer Engineering, 2019: 718-722.
27	SNELL J, SWERSKY K, ZEMEL R. Prototypical networks for few-shot learning[C]//Proc. of the Neural Information Processing Systems, 2017: 4077-4087.
28	HUANG H H, ZHANG F, ZHOU Y S, et al. High resolution SAR image synthesis with hierarchical generative adversarial networks[C]//Proc. of the IEEE International Geoscience and Remote Sensing Symposium, 2019: 2782-2785.
29	LI J W, QU C W, SHAO J Q. Ship detection in SAR images based on an improved faster R-CNN[C]//Proc. of the SAR in Big Data Era: Models, Methods and Applications, 2017.
30	TIAN Z, SHEN C H, CHEN H, et al. FCOS: fully convolutional one-stage object detection[C]//Proc. of the IEEE/CVF International Conference on Computer Vision, 2019: 9626-9635.
31	ZHOU X Y, WANG D Q, KRÄHENBÜHL P. Objects as points[EB/OL]. [2021-5-10]. https://arxiv.org/abs/1904.07850v2.
32	NICOLAS C, FRANCISCO M, GABRIEL S, et al. End-to-end object detection with transformers[C]//Proc. of the European Conference on Computer Vision, 2020: 213-229.
33	ASHISH V, NOAM S, NIKI P, et al. Attention is all you need[C]// Proc. of the 31st International Conference on Neural Information Processing Systems, 2017: 6000-6010.
34	WANG Y Y , WANG C , ZHANG H , et al. A SAR dataset of ship detection for deep learning under complex backgrounds[J]. Remote Sensing, 2019, 11 (7): 765. doi: 10.3390/rs11070765
35	WEI S J , ZENG X F , QU Q Z , et al. HRSID: a high-resolution SAR images dataset for ship detection and instance segmentation[J]. IEEE Access, 2020, 8, 120234- 120254. doi: 10.1109/ACCESS.2020.3005861
36	MINAEE S , BOYKOV Y Y , PORIKLI F , et al. Image segmentation using deep learning: a survey[J]. IEEE Trans. on Pattern Analysis and Machine Intelligence, 2021, doi: 10.1109/TPAMI.2021.3059968
37	SHELHAMER E , LONG J , DARRELL T . Fully convolutional networks for semantic segmentation[J]. IEEE Trans. on Pattern Analysis and Machine Intelligence, 2017, 39 (4): 640- 651. doi: 10.1109/TPAMI.2016.2572683
38	RONNEBERGER O, FISCHER P, BROX T. U-Net: convolutional networks for biomedicalimage segmentation[C]//Proc. of the Medical Image Computing and Computer-Assisted Intervention, 2015: 234-241.
39	BADRINARAYANAN V , KENDALL A , CIPOLLA R . SegNet: a deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE Trans. on Pattern Analysis and Machine Intelligence, 2017, 39 (12): 2481- 2495. doi: 10.1109/TPAMI.2016.2644615
40	CHEN L , PAPANDREOU G , KOKKINOS I , et al. DeepLab: semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs[J]. IEEE Trans. on Pattern Analysis and Machine Intelligence, 2018, 40 (4): 834- 848. doi: 10.1109/TPAMI.2017.2699184
41	MILLETARI F, NAVAB N, AHMADI S. V-Net: fully convolutional neural networks for volumetric medical image segmentation[C]//Proc. of the 4th International Conference on 3D Vision, 2016: 565-571.

类别	CE	Dice	MIoU	MPA
背景	✔	✘	99.85	99.92
背景	✔	✔	99.84	99.89
舰船目标	✔	✘	67.91	81.61
舰船目标	✔	✔	69.05	88.91
总计	✔	✘	83.88	90.76
总计	✔	✔	84.45	94.40

模型	Dice	AP	AP_0.5	AP_0.75	AP_s	AP_m	AP_l
Faster R-CNN	-	0.529	0.806	0.590	0.541	0.570	0.215
YOLOv4	-	0.376	0.788	0.324	0.390	0.367	0.088
UNet-S	✘	0.460	0.660	0.516	0.458	0.503	0.232
UNet-S	✔	0.544	0.756	0.607	0.547	0.560	0.319

模型	Dice	Size/MB	AR	AR_s	AR_m	AR_l
Faster R-CNN	-	-	0.595	0.581	0.716	0.521
YOLOv4	-	244.4	0.438	0.434	0.495	0.233
UNet-S	✘	94.9	0.643	0.640	0.683	0.517
UNet-S	✔	94.9	0.688	0.683	0.744	0.566

[1]	宋爽, 张悦, 张琳娜, 岑翼刚, 李浥东. 基于深度学习的轻量化目标检测算法[J]. 系统工程与电子技术, 2022, 44(9): 2716-2725.
[2]	韩啸, 陈世文, 陈蒙, 杨锦程. 基于互易点学习的LPI信号开集识别[J]. 系统工程与电子技术, 2022, 44(9): 2752-2759.
[3]	苗添, 曾虹程, 王贺, 陈杰. 基于迭代阈值分割的星载SAR洪水区域快速提取[J]. 系统工程与电子技术, 2022, 44(9): 2760-2768.
[4]	张涛, 杨小冈, 卢瑞涛, 谢学立, 刘闯. 基于关键点的遥感图像舰船目标检测[J]. 系统工程与电子技术, 2022, 44(8): 2437-2447.
[5]	王彩云, 吴钇达, 王佳宁, 马璐, 赵焕玥. 基于改进的CNN和数据增强的SAR目标识别[J]. 系统工程与电子技术, 2022, 44(8): 2483-2487.
[6]	张立民, 谭凯文, 闫文君, 张聿远. 基于多级跳线残差网络的雷达辐射源识别[J]. 系统工程与电子技术, 2022, 44(7): 2148-2156.
[7]	金国栋, 薛远亮, 谭力宁, 许剑锟. 基于孪生神经网络的目标跟踪算法进展研究[J]. 系统工程与电子技术, 2022, 44(6): 1805-1822.
[8]	傅东宁, 廖桂生, 黄岩, 张邦杰, 王幸. 基于图拉普拉斯嵌入的合成孔径雷达时变窄带干扰抑制算法[J]. 系统工程与电子技术, 2022, 44(6): 1846-1853.
[9]	盖明慧, 张苏, 孙卫天, 倪育德, 杨磊. 复数兼容全变分SAR目标结构特征增强[J]. 系统工程与电子技术, 2022, 44(6): 1862-1872.
[10]	徐安林, 张毓, 周峰. 基于Beta过程的高分辨ISAR成像[J]. 系统工程与电子技术, 2022, 44(6): 1873-1879.
[11]	韦娟, 杨皇卫, 宁方立. 基于NMF与CNN联合优化的声学场景分类[J]. 系统工程与电子技术, 2022, 44(5): 1433-1438.
[12]	赵晓枫, 徐叶斌, 吴飞, 牛家辉, 蔡伟, 张志利. 基于全局感知机制的地面红外目标检测方法[J]. 系统工程与电子技术, 2022, 44(5): 1461-1467.
[13]	纪朋徽, 代大海, 邢世其, 冯德军. 密集虚假运动目标生成方法[J]. 系统工程与电子技术, 2022, 44(5): 1502-1511.
[14]	邹虹, 白陈阳, 何鹏, 崔亚平, 王汝言, 吴大鹏. 基于分布式深度学习的边缘服务放置策略[J]. 系统工程与电子技术, 2022, 44(5): 1728-1737.
[15]	刘丰恺, 黄大荣, 郭新荣, 冯存前. 基于吕氏分布的机动目标参数化平动补偿方法[J]. 系统工程与电子技术, 2022, 44(4): 1166-1173.