强化散射特征的机载SAR实传图像盲超分辨重建

doi:10.12305/j.issn.1001-506X.2025.03.08

摘要/Abstract

摘要：

现有超分辨方法大多基于理想退化模型且易导致强散射信息均值化, 不适用于机载合成孔径雷达(synthetic aperture radar, SAR)实传图像的超分辨重建。针对该问题, 在机载SAR实传图像和精细成像图像之间建立了一个盲超分辨重建网络。首先, 采用生成对抗网络学习两个图像域之间的映射关系。其次, 通过注意力机制引导网络关注强散射区域。然后, 利用感知循环一致性损失保留图像纹理特征。最后, 在实测机载SAR数据集上验证了算法有效性, 重建结果的人类视觉系统信噪比和辐射分辨率分别提升了约30%和20%。特征分析及可视化表明, 所提方法提高了图像质量且重建出清晰的强散射特征。

关键词: 机载合成孔径雷达, 实传图像, 强化散射特征, 盲超分辨重建, 生成对抗网络

Abstract:

Most of the existing super-resolution methods are based on the ideal degradation models and tend to lead to the averaging effect of strong scattering information, which is not suitable for the super-resolution reconstruction of airborne synthetic aperture radar (SAR) real-time transmission images. To address this issue, a blind super-resolution reconstruction network is constructed between airborne SAR real-time transmission images and fine imaging images. Firstly, a generative adversarial network is used to learn the mapping relationship between two image domains. Secondly, an attention mechanism is used to guide the network to focus on strong scattering areas. Then, perceptual cycle consistency loss is utilized to further preserve the image texture features. Finally, the effectiveness of the algorithm is verified on the measured airborne SAR datasets, and the human visual system signal-to-noise ratio and radiation resolution of the reconstructed results are improved by about 30% and 20%, respectively. Feature analysis and visualization indicate that the proposed method can improve image quality and reconstruct clear strong scattering features.

Key words: airborne synthetic aperture radar (SAR), real-time transmission image, enhanced scattering feature, blind super-resolution reconstruction, generative adversarial network (GAN)

中图分类号:

TN957.52

贾钰嘉, 张思乾, 唐涛, 匡纲要. 强化散射特征的机载SAR实传图像盲超分辨重建[J]. 系统工程与电子技术, 2025, 47(3): 753-767.

Yujia JIA, Siqian ZHANG, Tao TANG, Gangyao KUANG. Blind super-resolution reconstruction of airborne SAR real-time transmission images with enhanced scattering features[J]. Systems Engineering and Electronics, 2025, 47(3): 753-767.

图/表 12

图1

图2

图3

图4

图5

图6

表1

图7

表2

表3

表4

图8

参考文献 37

1	RAN L , LIU Z , LI T , et al. Extension of map-drift algorithm for highly squinted SAR autofocus[J]. IEEE Journal of Selected Topics in Applied Earth Observations & Remote Sensing, 2017, 10 (9): 4032- 4044.
2	ZHAO L , XIE T , PERRIE W , et al. Detection of sea surface temperature fronts from SAR images[J]. Journal of Atmospheric and Oceanic Technology, 2022, 39 (12): 1919- 1926. doi: 10.1175/JTECH-D-22-0009.1
3	SINGH G , VENKATARAMAN G , YAMAGUCHI Y , et al. Capability assessment of fully polarimetric ALOS PALSAR data for discriminating wet snow from other scattering types in mountainous regions[J]. IEEE Trans.on Geoscience and Remote Sensing, 2014, 52 (2): 1177- 1196. doi: 10.1109/TGRS.2013.2248369
4	CAO C J , CUI Z Y , WANG L Y , et al. Cost-sensitive awareness-based SAR automatic target recognition for imbalanced data[J]. IEEE Trans.on Geoscience and Remote Sensing, 2022, 60, 5205316.
5	曲长文, 何友, 龚沈光. 机载SAR发展概况[J]. 现代雷达, 2002, 24 (1): 1-10, 14.
	QU C W , HE Y , GONG S G . Overview of airborne SAR deve-lopment[J]. Modern Radar, 2002, 24 (1): 1-10, 14.
6	张玉玲, 王鹏, 曲长文. 微型SAR发展状况[J]. 舰船电子对抗, 2008, 31 (5): 51- 55.
	ZHANG Y L , WANG P , QU C W . Development situation of miniature SAR[J]. Shipboard Electronic Countermeasure, 2008, 31 (5): 51- 55.
7	PARKER J A , KENYON R V , TROXEL D E . Comparison of interpolating methods for image resampling[J]. IEEE Trans.on Medical Imaging, 1983, 2 (1): 31- 39. doi: 10.1109/TMI.1983.4307610
8	SHEPPARD D G , PANCHAPAKESAN K , BILGIN A , et al. Lapped nonlinear interpolative vector quantizatoin and image super-resolution[J]. IEEE Trans.on Image Processing, 2000, 9 (2): 295- 298. doi: 10.1109/83.821746
9	MICHAEL E , ARIE F . Super-resolution reconstruction of image sequences[J]. IEEE Trans.on Pattern Analysis and Machine Intelligence, 1999, 21 (9): 817- 834. doi: 10.1109/34.790425
10	SCHULTZ R R, STEVENSON R L. Improved definition video frame enhancement[C]//Proc. of the International Conference on Acoustics, Speech, and Signal Processing, 1995: 2169-2172.
11	FREEMAN W T , JONES T R , PASZTOR E C . Example-based super-resolution[J]. IEEE Computer Graphics & Applications, 2002, 22 (2): 56- 65.
12	YANG J C , WRIGHT J , HUANG T S , et al. Image super-resolution via sparse representation[J]. IEEE Trans.on Image Processing, 2010, 19 (11): 2861- 2873.
13	XIAO G Y, ZHANG L. SAR image super-resolution reconstruction based on full-resolution discrimination[C]//Proc. of the IEEE International Conference on Image Processing, 2022: 691-695.
14	MOUSA A , BADRAN Y , SALAMA G , et al. Regression layer-based convolution neural network for synthetic aperture radar images: denoising and super-resolution[J]. The Visual Computer, 2023, 39 (4): 1295- 1306.
15	闵锐, 杨学志, 董张玉, 等. 结构增强型生成对抗网络SAR图像超分辨率重建[J]. 地理与地理信息科学, 2021, 37 (2): 47- 53.
	MIN R , YANG X Z , DONG Z Y , et al. Structure enhanced generative adversarial network SAR image super-resolution reconstruction[J]. Geography and Geo-Information Science, 2021, 37 (2): 47- 53.
16	SHEN H F , LIN L P , LI J , et al. A residual convolutional neural network for polarimetric SAR image super-resolution[J]. ISPRS Journal of Photogrammetry & Remote Sensing, 2020, 161, 90- 108.
17	WU Z R , ZHAO Z Y , MA P F , et al. Real-world DEM super-resolution based on generative adversarial networks for improving InSAR topographic phase simulation[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2021, 14, 8373- 8385.
18	WU W F , HUANG X , SHAO Z F , et al. SAR-DRDNet: a SAR image despeckling network with detail recovery[J]. Neurocomputing, 2022, 493, 253- 267.
19	LIU S Q , LEI Y , ZHANG L Y , et al. MRDDANet: a multiscale residual dense dual attention network for SAR image denoising[J]. IEEE Trans.on Geoscience and Remote Sensing, 2022, 60, 5214213.
20	ZHANG K, LIANG J Y, VAN G L, et al. Designing a practical degradation model for deep blind image super-resolution[C]//Proc. of the IEEE/CVF International Conference on Computer Vision, 2021: 4771-4780.
21	WANG X T, XIE L B, DONG C, et al. Real-ESRGAN: training real-world blind super-resolution with pure synthetic data[C]//Proc. of the IEEE/CVF International Conference on Computer Vision Workshops, 2021: 1905-1914.
22	SHOCKER A, COHEN N, IRANI M. Zero-shot super-resolution using deep internal learning[C]//Proc. of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018: 3118-3126.
23	CHANG T W, CHIU W C, HUANG C C. Find the way back: invertible kernel estimator for blind image super-resolution[C]//Proc. of the IEEE International Conference on Acoustics, Speech and Signal Processing, 2022: 2145-2149.
24	LIU A R , LIU Y H , GU J J , et al. Blind image super-resolution: a survey and beyond[J]. IEEE Trans.on Pattern Analysis and Machine Intelligence, 2023, 45 (5): 5461- 5480.
25	YUAN Y, LIU S Y, ZHANG J W, et al. Unsupervised image super-resolution using cycle-in-cycle generative adversarial networks[C]//Proc. of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2018.
26	GOODFELLOW I J, POUGET-ABADIE J, MIRZA M. Gene-rative adversarial nets[C]//Proc. of the 27th International Conference on Neural Information Processing Systems, 2014: 2672-2680.
27	易拓源, 户盼鹤, 刘振. 基于改进CycleGAN的ISAR图像超分辨方法[J]. 信号处理, 2023, 39 (2): 323- 334.
	YI T Y , HU P H , LIU Z . ISAR images super-resolution method based on ameliorated CycleGAN[J]. Journal of Signal Processing, 2023, 39 (2): 323- 334.
28	ZHU J Y, PARK T, ISO-LA P, et al. Unpaired image-to-image translation using cycle-consistent adversarial networks[C]//Proc. of the IEEE International Conference on Computer Vision, 2017.
29	AO D Y , DUMITRU C O , SCHWARZ G , et al. Dialectical GAN for SAR image translation: from Sentinel-1 to TerraSAR-X[J]. Remote Sensing, 2018, 10 (10): 1597.
30	李萌, 刘畅. 基于特征复用的膨胀-残差网络的SAR图像超分辨重建[J]. 雷达学报, 2020, 9 (2): 363- 372.
	LI M , LIU C . Super-resolution reconstruction of SAR images based on feature reuse dilated-residual convolutional neural networks[J]. Journal of Radars, 2020, 9 (2): 363- 372.
31	LUO Z Y , YU J P , LIU Z H . The super-resolution reconstruction of SAR image based on the improved FSRCNN[J]. The Journal of Engineering, 2019, 2019 (19): 5975- 5978.
32	YANG W, MA Z Q, SHI Y R. SAR image super-resolution based on artificial intelligence[C]//Proc. of the IEEE International Geoscience and Remote Sensing Symposium, 2022.
33	WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]//Proc. of the 15th European Conference on Computer Vision, 2018.
34	ZHANG W L, LIU Y H, DONG C, et al. RankSRGAN: gene-rative adversarial networks with ranker for image super-reso-lution[C]//Proc. of the IEEE/CVF International Conference on Computer Vision, 2019.
35	BLAU Y, MECHREZ R, TIMOFTE R, et al. The 2018 PIRM challenge on perceptual image super-resolution[C]//Proc. of the European Conference on Computer Vision, 2018.
36	LIU Y J, YU Z, LI C S. A novel quality evaluation algorithm for SAR image based on human visual system[C]//Proc. of the IEEE International Geoscience and Remote Sensing Sympo-sium, 2013.
37	SELVARAJU R R, COGSWELL M, DAS A, et al. Grad-CAM: visual explanations from deep networks via gradient-based localization[C]//Proc. of the IEEE International Conference on Computer Vision, 2017.

图像编号	评估指标	低分辨率图像	高分辨率图像	超分辨重建图像	提升百分比/%
1	HVSNR	4.108 0	6.485 4	5.692 4	38.57
1	γ/dB	3.594 4	3.191 8	3.258 1	9.36
2	HVSNR	4.843 3	6.632 2	6.225 3	28.53
2	γ/dB	5.136 2	3.720 0	3.854 8	24.95
3	HVSNR	5.499 2	7.777 2	7.035 2	27.93
3	γ/dB	4.736 5	3.687 7	3.673 9	22.43

切片编号	数据类型	全局指标		局部指标
切片编号	数据类型	μ	σ	n	v
1	高分辨率	54.63	58.09	730	249.62
	超分辨	54.09	52.32	700	235.27
	相对误差/%	0.99	9.93	4.11	5.75
2	高分辨率	35.96	27.36	150	228.09
	超分辨	34.41	29.25	186	227.68
	相对误差/%	4.31	6.91	24	0.18
3	高分辨率	60.71	41.80	620	227.59
	超分辨	77.96	43.52	698	230.90
	相对误差/%	28.41	4.11	12.58	1.45
4	高分辨率	50.05	38.29	432	236.20
	超分辨	51.72	37.69	396	230.78
	相对误差/%	3.34	1.57	14.58	2.29
5	高分辨率	56.68	56.12	720	241.75
	超分辨	61.53	48.72	631	232.27
	相对误差/%	8.56	13.19	12.36	3.92
6	高分辨率	35.72	20.06	38	229.61
	超分辨	39.56	20.98	43	227.58
	相对误差/%	10.75	4.54	13.16	0.88

算法	参数量/M	GFLOPs
CycleGAN^[28]	11.37	56.04
本文方法	11.54	56.06

实验方法	HVSNR	γ/dB	全局指标/%		局部指标/%
实验方法	HVSNR	γ/dB	μ	σ	n	v
高分辨率	2.168 9	3.334 4	-	-	-	-
低分辨率	1.076 6	4.313 1	115.03	46.78	285.07	2.04
CycleGAN^[28]	1.735 3	4.048 4	18.67	18.49	41.09	7.96
CycleGAN+L_per(·)	1.879 1	3.935 7	20.27	14.13	39.55	4.95
CycleGAN+CBAM	1.824 2	3.871 7	21.27	12.62	34.54	3.87
本文方法	1.954 6	3.775 1	17.53	12.81	30.51	3.52

[1]	张兰, 张彪, 梁天一, 朱辉杰. 面向电磁信息智能控制的生成对抗网络研究进展[J]. 系统工程与电子技术, 2025, 47(3): 730-744.
[2]	薛丽莎, 葛瑞星, 朱宇轩, 鲍雁飞. 基于SABEGAN的通信干扰信号生成与效能分析[J]. 系统工程与电子技术, 2024, 46(6): 2155-2163.
[3]	汤寓麟, 王黎明, 余德荧, 李厚朴, 刘敏, 张卫东. 基于CSLS-CycleGAN的侧扫声纳水下目标图像样本扩增法[J]. 系统工程与电子技术, 2024, 46(5): 1514-1524.
[4]	马兰, 孟诗君, 吴志军. 基于BERT与生成对抗的民航陆空通话意图挖掘[J]. 系统工程与电子技术, 2024, 46(2): 740-750.
[5]	庞世燕, 郝京京, 邢立宁, 谭旭. 融合域自适应和弱监督策略的遥感影像建筑物提取算法[J]. 系统工程与电子技术, 2023, 45(6): 1616-1623.
[6]	陈丽, 方梓涵, 梅立泉. 基于GAN的直扩信号生成算法[J]. 系统工程与电子技术, 2023, 45(5): 1544-1552.
[7]	曾浩然, 冯蕴雯, 路成, 潘维煌. 基于改进生成对抗网络的国产民机航材消耗预测方法[J]. 系统工程与电子技术, 2023, 45(10): 3132-3138.
[8]	秦博伟, 蒋磊, 许华, 牛伟宇. 基于RE-GAN的调制信号开集识别算法[J]. 系统工程与电子技术, 2023, 45(10): 3321-3328.
[9]	刘戎翔, 吴琳, 谢智歌, 刘虹麟. 基于生成对抗网络的防空体系态势辅助分析[J]. 系统工程与电子技术, 2022, 44(8): 2522-2529.
[10]	徐平亮, 崔亚奇, 熊伟, 熊振宇, 顾祥岐. 生成式中断航迹接续关联方法[J]. 系统工程与电子技术, 2022, 44(5): 1543-1552.
[11]	曹鹏宇, 杨承志, 石礼盟, 吴宏超. 基于PSO-DBSCAN和SCGAN的未知雷达信号处理方法[J]. 系统工程与电子技术, 2022, 44(4): 1158-1165.
[12]	邵凯, 朱苗苗, 王光宇. 基于生成对抗与卷积神经网络的调制识别方法[J]. 系统工程与电子技术, 2022, 44(3): 1036-1043.
[13]	曹鹏宇, 杨承志, 石礼盟, 吴宏超. 基于DAE-GAN网络的LPI雷达信号增强[J]. 系统工程与电子技术, 2021, 43(9): 2493-2500.
[14]	赵凡, 金虎. 基于GAN的通信干扰波形生成技术[J]. 系统工程与电子技术, 2021, 43(4): 1080-1088.
[15]	田相轩, 石志强. 基于改进型生成对抗网络的指挥信息系统模拟数据生成算法[J]. 系统工程与电子技术, 2021, 43(1): 163-170.