SAR图像目标轮廓增强预处理模块设计

doi:10.12305/j.issn.1001-506X.2024.12.09

摘要/Abstract

摘要：

针对合成孔径雷达(synthetic aperture radar，SAR)图像中的3个通道数据均相同、以至于在使用基于深度学习的目标检测网络对其进行目标检测时会造成通道信息利用效率低下的问题，基于对SAR图像中各种目标物轮廓特点的研究，提出一种基于平滑和锐化滤波的通道扩展预处理算法模块，并将其命名为ORLM (Original, Roberts, Laplace, Mean)模块。所提算法可被封装、集成并应用于目标检测算法的数据读取程序，可将每个通道数据均相同的SAR图像进行通道扩展，并且保证扩展后的通道数据充分包含目标的全部轮廓信息。通过将有否搭配本预处理算法的目标检测网络在不同舰船目标检测数据集上进行训练、测试及对比实验，实验结果表明，所提预处理算法可被应用在各种目标检测算法中，起到提升检测准确性的作用，且不会明显降低检测实时性。

关键词: 图像处理, 合成孔径雷达, 深度学习, 目标检测

Abstract:

In response to the issue that the every channel's data in three channels of synthetic aperture radar (SAR) images is the same which may cause channel information redundancy when deep learning-based target detection network detects the targets of SAR, a channel expansion preprocessing algorithm module based on smoothing and sharpening filtering is proposed, which is then named as ORLM (Original, Roberts, Laplace, Mean) block. The proposed algorithm in this article can be encapsulated, integrated, and applied to the data reading program of the target detection algorithm. It can extend the channel of SAR images with the same data in each channel, and ensure that the expanded channel data fully contains the contour information of the target. Through training, testing and comparative experiments of the target detection network with and without the proposed preprocessing algorithm on different ship target detection datasets, the experimental results show that the preprocessing algorithm proposed can be applied to various target detection algorithms and can improve detection accuracy without significantly reducing of the real-time detection performance.

Key words: image processing, synthetic aperture radar (SAR), deep learning, target detection

中图分类号:

TP391

龚峻扬, 付卫红, 刘乃安. SAR图像目标轮廓增强预处理模块设计[J]. 系统工程与电子技术, 2024, 46(12): 4010-4017.

Junyang GONG, Weihong FU, Naian LIU. Design of SAR image target contour enhancement preprocessing module[J]. Systems Engineering and Electronics, 2024, 46(12): 4010-4017.

图/表 10

图1

图2

图3

图4

表1

表2

图5

表3

表4

表5

参考文献 31

1	张冬冬, 王春平, 付强. 基于特征增强网络的SAR图像舰船目标检测[J]. 系统工程与电子技术, 2023, 45 (4): 1032- 1039. doi: 10.12305/j.issn.1001-506X.2023.04.12
	ZHANG D D , WANG C P , FU Q . Ship target detection in SAR image based on feature-enhanced network[J]. Systems Engineering and Electronics, 2023, 45 (4): 1032- 1039. doi: 10.12305/j.issn.1001-506X.2023.04.12
2	杨磊, 夏亚波, 廖仙华, 等. 双层稀疏贝叶斯学习ISAR超分辨成像算法[J]. 系统工程与电子技术, 2023, 45 (5): 1371- 1379. doi: 10.12305/j.issn.1001-506X.2023.05.13
	YANG L , XIA Y B , LIAO X H , et al. Super-resolution ISAR imagery algorithm based on bi-sparsity Bayesian learning[J]. Systems Engineering and Electronics, 2023, 45 (4): 1371- 1379. doi: 10.12305/j.issn.1001-506X.2023.05.13
3	杨宇超, 方明, 赵晨帆, 等. 高速机动目标长时间相参积累算法[J]. 系统工程与电子技术, 2023, 45 (5): 1359- 1370. doi: 10.12305/j.issn.1001-506X.2023.05.12
	YANG Y C , FANG M , ZHAO C F , et al. Long-time coherent integration algorithm for high-speed maneuvering targets[J]. Systems Engineering and Electronics, 2023, 45 (5): 1359- 1370. doi: 10.12305/j.issn.1001-506X.2023.05.12
4	陈涛, 刘福悦, 李金鑫, 等. 基于深度分割的端到端雷达信号分选[J]. 系统工程与电子技术, 2023, 45 (5): 1351- 1358. doi: 10.12305/j.issn.1001-506X.2023.05.11
	CHEN T , XIU F Y , LI J X , et al. End-to-end radar signal sorting based on deep segmentation[J]. Systems Engineering and Electronics, 2023, 45 (5): 1351- 1358. doi: 10.12305/j.issn.1001-506X.2023.05.11
5	苏娟, 杨龙, 黄华, 等. 用于SAR图像小目标舰船检测的改进SSD算法[J]. 系统工程与电子技术, 2020, 42 (5): 1026- 1034. doi: 10.3969/j.issn.1001-506X.2020.05.08
	SU J , YANG L , HUANG H , et al. Improved SSD algorithm for SAR image small target ship detection[J]. Systems Engineering and Electronics, 2020, 42 (5): 1026- 1034. doi: 10.3969/j.issn.1001-506X.2020.05.08
6	PAN X B , WANG W B , WU L , et al. Improved moving target imaging method for a multichannel HRWS SAR system[J]. IEEE Geoscience and Remote Sensing Letters, 2023, 20, 4008505.
7	KIM J H , SHIN S J , KIM S H , et al. EO-augmented building segmentation for airborne SAR imagery[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19, 4013005.
8	GUO Q , LIU J N , KALIUZHNYI M . YOLOX-SAR: high-precision object detection system based on visible and infrared sensors for SAR remote sensing[J]. IEEE Sensors Journal, 2022, 22 (17): 17243- 17253. doi: 10.1109/JSEN.2022.3186889
9	吕勤哲, 全英汇, 沙明辉, 等. 基于集成深度学习的有源干扰智能分类[J]. 系统工程与电子技术, 2022, 44 (12): 3595- 3602. doi: 10.12305/j.issn.1001-506X.2022.12.02
	LV Q Z , QUAN Y H , SHA M H , et al. Ensemble deep learning-based intelligent classification of active jamming[J]. Systems Engineering and Electronics, 2022, 44 (12): 3595- 3602. doi: 10.12305/j.issn.1001-506X.2022.12.02
10	樊成, 王布宏, 田继伟. 基于多任务学习图卷积模型的航空网络节点分类[J]. 系统工程与电子技术, 2022, 44 (7): 2341- 2349. doi: 10.12305/j.issn.1001-506X.2022.07.31
	FAN C , WANG B H , TIAN J W . Node classification of airline network based on the graph convolution network model with multi-task learning[J]. Systems Engineering and Electronics, 2022, 44 (7): 2341- 2349. doi: 10.12305/j.issn.1001-506X.2022.07.31
11	李瑞泽, 张双辉, 刘永祥. 基于卷积ADMM网络的高效结构化稀疏ISAR成像方法[J]. 系统工程与电子技术, 2023, 45 (1): 56- 70. doi: 10.12305/j.issn.1001-506X.2023.01.08
	LI R Z , ZHANG S H , LIU Y X . Computational efficient structural sparse ISAR imaging method based on convolutional ADMM-net[J]. Systems Engineering and Electronics, 2023, 45 (1): 56- 70. doi: 10.12305/j.issn.1001-506X.2023.01.08
12	李睿峰, 许爱强, 孙伟超, 等. 基于元学习的航空电子设备特征选择算法推荐方法[J]. 系统工程与电子技术, 2021, 43 (7): 2011- 2020. doi: 10.12305/j.issn.1001-506X.2021.07.34
	LI R F , XU A Q , SUN W C , et al. Recommendation method for avionics feature selection algorithm basedon meta-learning[J]. Systems Engineering and Electronics, 2021, 43 (7): 2011- 2020. doi: 10.12305/j.issn.1001-506X.2021.07.34
13	周晓玲, 张朝霞, 鲁雅, 等. 基于改进R-FCN的SAR图像识别[J]. 系统工程与电子技术, 2022, 44 (4): 1202- 1209. doi: 10.12305/j.issn.1001-506X.2022.04.17
	ZHOU X L , ZHANG Z X , LU Y , et al. SAR image recognition based on improved R-FCN[J]. Systems Engineering and Electronics, 2022, 44 (4): 1202- 1209. doi: 10.12305/j.issn.1001-506X.2022.04.17
14	何楚, 张宇, 廖紫纤, 等. 基于压缩感知的SAR图像CFAR目标检测算法[J]. 武汉大学学报(信息科学版), 2014, 39 (7): 878- 882.
	HE C , ZHANG Y , LIAO Z X , et al. SAR image CFAR target detection algorithm based on compressed sensing[J]. Geomatics and Information Science of Wuhan University, 2014, 39 (7): 878- 882.
15	ZHOU J , XIE J H . An improved quantile estimator with its application in CFAR detection[J]. IEEE Geoscience and Remote Sensing Letters, 2023, 20, 3507705.
16	ROSU F . Dimension compressed CFAR for massive MIMO radar[J]. IEEE Geoscience and Remote Sensing Letters, 2023, 20, 3504305.
17	ZHOU Y , HU H , ZHAO J Q , et al. Few-shot object detection via context-aware aggregation for remote sensing images[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19, 6509605.
18	LIU D Y , ZHANG J P , LI T , et al. A lightweight object detection and recognition method based on light global-local module for remote sensing images[J]. IEEE Geoscience and Remote Sensing Letters, 2023, 20, 6007105.
19	YUAN Y , ZHANG Y L . OLCN: an optimized low coupling network for small objects detection[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19, 8022005.
20	孙显, 王智睿, 孙元睿, 等. AIR-SARShip-1.0:高分辨率SAR舰船检测数据集[J]. 雷达学报, 2019, 8 (6): 852- 862.
	SUN X , WANG Z R , SUN Y R , et al. AIR-SARShip-1.0: high-resolution SAR ship detection dataset[J]. Journal of Radars, 2019, 8 (6): 852- 862.
21	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.
22	ZHANG T W , ZHANG X L , KE X , et al. LS-SSDD-v1. 0: a deep learning dataset dedicated to small ship detection from large-scale Sentinel-1 SAR images[J]. Remote Sensing, 2020, 12 (18): 2997.
23	龚峻扬, 付卫红, 方厚章. SAR图像舰船目标检测的轻量化和特征增强研究[J]. 西安电子科技大学学报, 2024, 51 (2): 96- 106.
	GONG J Y , FU W H , FANG H Z . Research on lightweight and feature enhancement of SAR image ship targets detection[J]. Journal of Xidian University, 2024, 51 (2): 96- 106.
24	徐从安, 苏航, 李健伟, 等. RSDD-SAR: SAR舰船斜框检测数据集[J]. 雷达学报, 2022, 11 (4): 581- 599.
	XU C A , SU H , LI J W , et al. RSDD-SAR: rotated ship detection dataset in SAR images[J]. Journal of Radars, 2022, 11 (4): 581- 599.
25	仇晓兰, 焦泽坤, 彭凌霄, 等. SARMV3D-1.0: SAR微波视觉三维成像数据集[J]. 雷达学报, 2021, 10 (4): 485- 498.
	QIU X L , JIAO Z K , PENG L X , et al. SARMV3D-1.0: synthetic aperture radar microwave vision 3D imaging dataset[J]. Journal of Radars, 2021, 10 (4): 485- 498.
26	JIANG J H , FU X J , QIN R , et al. High-speed lightweight ship detection algorithm based on YOLO-v4 for three-channels RGB SAR image[J]. Remote Sensing, 2021, 13 (10): 1909.
27	LAN Z Z, LIN M, LI X C, et al. Beyond Gaussian pyramid: multi-skip feature stacking for action recognition[C]//Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2015: 204-212.
28	WANG L M , ZHOU K , CHU A L , et al. An improved light-weight traffic sign recognition algorithm based on YOLOv4-tiny[J]. IEEE Access, 2021, 9, 124963- 124971.
29	WANG P H , WANG X F , LIU Y F , et al. Research on road object detection model based on YOLOv4 of autonomous vehicle[J]. IEEE Access, 2024, 12, 8198- 8206.
30	MA N, ZHANG X, ZHENG H T, et al. Shufflenet v2: practical guidelines for efficient CNN architecture design[C]// Proc. of the European Conference on Computer Vision, 2018: 116-131.
31	HOWARD A, SANDLER M, CHU G, et al. Searching for mobilenetv3[C]//Proc. of the IEEE CVF International Conference on Computer Vision, 2019: 1314-1324.

传感器	极化方式	船舶	海洋环境	分辨率/m	船舶位置	比例
RadarSat-2 TerraSAR-X Sentinel-1	HH, VV, VH, HV	不同尺寸和材料	良好和不良状况	1~5	海里和离岸	1:1 1:2 2:1

	算法	mAP@ 0.5	mAP@ 0.5:0.95	F1_ Score
锐化+平滑	ord+Roberts(ord)+Gaussian(ord)	0.977	0.757	0.966
	ord+Roberts(ord)+Median(ord)	0.980	0.756	0.972
	ord+Roberts(ord)+Mean(ord)	0.981	0.757	0.972
锐化+平滑+锐化	ord+Roberts(ord)+Roberts(Gaussian(ord))	0.98	0.749	0.973
	ord+Roberts(ord)+Laplace(Gaussian(ord))	0.978	0.748	0.970
	ord+Roberts(ord)+Sobel(Gaussian(ord))	0.979	0.762	0.969
	ord+Roberts(ord)+Roberts(Median(ord))	0.979	0.747	0.972
	ord+Roberts(ord)+Laplace(Median(ord))	0.981	0.75	0.973
	ord+Roberts(ord)+Sobel(Median(ord))	0.974	0.754	0.965
	ord+Roberts(ord)+Roberts(Mean(ord))	0.979	0.751	0.971
	ord+Roberts(ord)+Laplace(Mean(ord))	0.983	0.758	0.974
	ord+Roberts(ord)+Sobel(Mean(ord))	0.979	0.762	0.969
	ord+ord+ord	0.981	0.748	0.970

网络	mAP@0.5	mAP@0.5:0.95	F1_Score	FLOPs	FPS
CSPDarkNet	0.981	0.747	0.970	50.6	146
CSPDarkNet+ORLM模块	0.981	0.761	0.974	50.6	136
ShuffleNetV2	0.969	0.704	0.950	4.3	221
ShuffleNetV2+ORLM模块	0.969	0.716	0.943	4.3	221
MobileNetV3	0.952	0.664	0.94	1.0	309
MobileNetV3+ORLM模块	0.951	0.678	0.938	1.0	300
CSPDarkNet53	0.960	0.726	0.949	16.9	188
CSPDarkNet53+ORLM模块	0.965	0.733	0.953	16.9	173

网络	mAP@0.5	mAP@0.5:0.95	F1_Score	FLOPs	FPS
CSPDarkNet	0.837	0.392	0.802	50.6	84
CSPDarkNet+ORLM模块	0.835	0.400	0.813	50.6	124
ShuffleNetV2	0.832	0.386	0.797	4.3	256
ShuffleNetV2+ORLM模块	0.842	0.390	0.795	4.3	425
MobileNetV3	0.820	0.375	0.780	1.0	300
MobileNetV3+ORLM模块	0.833	0.388	0.779	1.0	450
CSPDarkNet53	0.825	0.310	0.792	16.9	99
CSPDarkNet53+ORLM模块	0.830	0.318	0.799	16.9	130

算法	mAP@0.5	mAP@0.5:0.95	F1_Score	FLOPs	FPS
YOLOv5m+ORLM模块	0.981	0.761	0.974	50.6	136
YOLOv5m+NSLP	0.977	0.748	0.969	50.6	139