基于方差拟合插值的SAR图像数据集扩充方法

doi:10.12305/j.issn.1001-506X.2025.10.11

系统工程与电子技术 ›› 2025, Vol. 47 ›› Issue (10): 3235-3250.doi: 10.12305/j.issn.1001-506X.2025.10.11

• 传感器与信号处理 • 上一篇

基于方差拟合插值的SAR图像数据集扩充方法

李璇¹(), 贡琦¹(), 何姿²^,*(), 樊振宏²(), 丁大志²()

1. 南京理工大学电子工程与光电技术学院，江苏南京 210094
2. 南京理工大学微电子学院，江苏南京 210094

收稿日期:2024-07-01 出版日期:2025-10-25 发布日期:2025-10-23
通讯作者: 何姿 E-mail:lxuan0820@163.com;1441428166@qq.com;zihe@njust.edu.cn;zhfan@njust.edu.cn;dzding@njust.edu.cn
作者简介:李　璇（1992—），女，博士研究生，主要研究方向为计算电磁学及工程应用
贡　琦（1999—），女，硕士，主要研究方向为计算电磁学及工程应用
樊振宏（1978—），男，教授，博士，主要研究方向为计算电磁学、天线设计与优化
丁大志（1979—），男，教授，博士，主要研究方向为计算电磁学、微波成像
基金资助:
国家自然科学基金（62322107，62235006）资助课题

Variance-fitting interpolation-based method for SAR image dataset expansion

Xuan LI¹(), Qi GONG¹(), Zi HE²^,*(), Zhenhong FAN²(), Dazhi DING²()

1. School of Electronic and Optical Engineering，Nanjing University of Science and Technology，Nanjing 210094，China
2. School of Microelectronics，Nanjing University of Science and Technology，Nanjing 210094，China

Received:2024-07-01 Online:2025-10-25 Published:2025-10-23
Contact: Zi HE E-mail:lxuan0820@163.com;1441428166@qq.com;zihe@njust.edu.cn;zhfan@njust.edu.cn;dzding@njust.edu.cn

摘要/Abstract

摘要：

针对合成孔径雷达数据集稀缺，生成对抗网络图像生成时计算量高、训练不稳定和大数据依赖等问题，提出一种基于方差拟合的多方位合成孔径雷达图像插值方法。通过挖掘不同插值方法的特性差异，实现了水平面方位角数据扩充。实验结果表明，主极化插值效果显著优于交叉极化，均方根误差降低32.7%，峰值信噪比提升2.8 dB，即使在海情5级时，仍保持高保真度（径向积分相似度>0.994），验证了方法的有效性。该方法充分挖掘图像插值方法之间的差异性，为合成孔径雷达图像数据集的扩充提供了一种新的解决方案。

关键词: 数据集扩充, 图像仿真, 插值算法

Abstract:

In response to the scarcity of synthetic aperture radar （SAR） datasets, the high computational, unstable training, and dependence on big data in generating adversarial network images, this paper proposes a variance-adaptive multi-azimuth SAR image interpolation method that exploits the inherent characteristics of different interpolation approaches to achieve horizontal-plane azimuth data augmentation. Experimental results demonstrate that the interpolation performance for co-polarized channels significantly outperforms cross-polarized channels, with a 32.7% reduction in root mean square error and a 2.8 dB improvement in peak signal-to-noise ratio. Notably, the method maintains high fidelity （radial integral similarity>0.994） even under level-5 sea state conditions. The effectiveness of the method is validated. The proposed solution provides a novel approach for SAR dataset augmentation by effectively leveraging the differential features of interpolation techniques.

Key words: dataset expansion, image simulation, interpolation algorithm

中图分类号:

TP 751.1

李璇, 贡琦, 何姿, 樊振宏, 丁大志. 基于方差拟合插值的SAR图像数据集扩充方法[J]. 系统工程与电子技术, 2025, 47(10): 3235-3250.

Xuan LI, Qi GONG, Zi HE, Zhenhong FAN, Dazhi DING. Variance-fitting interpolation-based method for SAR image dataset expansion[J]. Systems Engineering and Electronics, 2025, 47(10): 3235-3250.

图/表 16

图1

图2

图3

图4

图5

图6

图7

图8

图9

图10

图11

图12

图13

表1

图14

图15

参考文献 30

1	NASROLLAHI K, MOESLUND T. B. Super-resolution: a comprehensive survey[J]. Machine Vision and Applications, 2014, 25 (6): 1423- 1468. doi: 10.1007/s00138-014-0623-4
2	林毓秀. 序列图像帧间插值方法及应用研究[D]. 济南: 山东财经大学, 2021.
	LIN Y X. Research on interpolation method and application of sequence Images[D]. Jinan: Shandong University of Finance and Economics, 2021.
3	ROHDE G K , ALDROUBI A , HEALY D M J. Interpolation artifacts in sub-pixel image registration[J]. IEEE Trans. on Image Processing, 2009, 18 (2): 333- 345.
4	盛晓艳, 龚超. 数字图像插值算法比较研究[J]. 电脑知识与技术, 2019, 15 (8): 153- 156.
	SHENG X Y, GONG C. Research on digital image interpolation algorithm[J]. Computer Knowledge and Technology, 2019, 15 (8): 153- 156.
5	张智邦, 李桂清, 韦国栋, 等. 形状插值算法综述[J]. 计算机辅助设计与图形学学报, 2015, 27 (8): 1376- 1387. doi: 10.3969/j.issn.1003-9775.2015.08.004
	ZHANG Z B, LI G Q, WEI G D, et al. Survey on shape interpolation[J]. Journal of Computer-aided Design and Computer Graphics, 2015, 27 (8): 1376- 1387. doi: 10.3969/j.issn.1003-9775.2015.08.004
6	GADERMAYR M, KOLLER L, TSCHUCHNIG M, et al. MixUp-MIL: a study on linear & multilinear interpolation-based data augmentation for whole slide image classification[EB/OL]. [2024-07-01].https://arxiv.org/abs/2311.03052.
7	LUMINI A . Comparison of different image data augmentation approaches[J]. Journal of Imaging, 2021, 7(12): 254.
8	ZITOVA B, FLUSSER J. Image registration methods: a survey[J]. Image and Vision Computing, 2003, 21 (11): 977- 1000. doi: 10.1016/S0262-8856(03)00137-9
9	梁墨翰. 基于生成对抗网络的多视角SAR图像目标数据扩充研究[D]. 成都: 电子科技大学, 2021.
	LIANG M H. Research on multi-view SAR image target data augmentation based on generative adversarial networks[D]. Chengdu: University of Electronic Science and Technology of China, 2021.
10	NALEPA J, MRUKWA G, PIECHACZEK S, et al. Data augmentation via image registration[C]// Proc. of the IEEE International Conference on Image Processing, 2019: 4250−4254.
11	MADASAMY S, DABU C N K, SUMATHI V, et al. A robust image registration using multi-scale feature extraction techniques using deep learning[C]// Proc. of the 4th International Conference on Smart Electronics and Communication, 2023: 1422−1427.
12	WANG C W, PEI J F, LIU X Y, et al. SAR target image generation method using azimuth-controllable generative adversarial network[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2022, 15, 9381- 9397. doi: 10.1109/JSTARS.2022.3218369
13	OH J, KIM M. PeaceGAN: a GAN-based multi-task learning method for SAR target image generation with a pose estimator and an auxiliary classifier[J]. Remote Sensing, 2021, 13 (19): 3939. doi: 10.3390/rs13193939
14	DU S Y, HONG J, WANG Y, et al. A high-quality multicategory SAR images generation method with multiconstraint GAN for ATR[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19, 4011005.
15	GAO S W, QIU S L, MA Z Y, et al. SVAE-WGAN-based soft sensor data supplement method for process industry[J]. IEEE Sensors Journal, 2022, 22, 601- 610.
16	WU D H, ZHANG W Y, ZHANG P F. DPBA-WGAN: a vector-valued differential private bilateral alternative scheme on WGAN for image generation[J]. IEEE Access, 2023, 11, 13889- 13905. doi: 10.1109/ACCESS.2023.3243473
17	ZHANG H Y, WANG Q, ZHANG R H, et al. Image reconstruction for electrical mpedance tomography （EIT） with improved wasserstein generative adversarial network （WGAN）[J]. IEEE Sensors Journal, 2023, 23 (5): 4466- 4475. doi: 10.1109/JSEN.2022.3197663
18	TAKAHASHI M, WATANABE H. Face image generation for illustration by WGAN-GP using landmark information[C]// Proc. of the IEEE 10th Global Conference on Consumer Electronics, 2021: 936−937.
19	HU F S, DONG C Y, WANG M S, et al. WGAN-GP with residual network model for lithium battery thermal image data expansion with quantitative metrics[C]// Proc. of the IEEE 6th International Electrical and Energy Conference, 2023: 4030−4035.
20	LIU X Q, LI G, ZHAO Z, et al. EAF-WGAN: enhanced alignment fusion-wasserstein generative adversarial network for turbulent image restoration[J]. IEEE Trans. on Circuits and Systems for Video Technology, 2023, 33 (10): 5605- 5616. doi: 10.1109/TCSVT.2023.3262685
21	YAN H, ZHANG Z W, XU J, et al. UW-CycleGAN: model-driven CycleGAN for underwater image restoration[J]. IEEE Trans. on Geoscience and Remote Sensing, 2023, 61
22	ZI Y, XIE F Y, SONG X D, et al. Thin cloud removal for remote sensing images using a physical-model-based CycleGAN with unpaired data[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19, 1004605.
23	GUZEL S, YAVUZ S. Infrared image generation from RGB images using CycleGAN[C]// Proc. of the International Conference on INnovations in Intelligent SysTems and Applications, 2022.
24	YU M T, YE X F, CHEN A, et al. Simulated sonar image generation method based on improved CycleGAN[C]// Proc. of the IEEE 11th International Conference on Computer Science and Network Technology, 2023: 281−285.
25	武士想, 尚鹏, 王立功. 小波-Lagrange方法进行医学图像层间插值[J]. 中国图象图形学报, 2016, 21(1): 78−85.
	WU S X, SHANG P, WANG L G. Inter-slice interpolation for medical images by using the wavelet-Lagrange method[J]. Journal of Image and Graphics, 2016, 21(1) : 78−85.
26	LI Z N, CHEN X H, GOU S N , et al. WavEnhancer: unifying wavelet and transformer for image enhancement[J]. Journal of Computer Science and Technology, 2024, 39(2): 336.
27	SAHA M, NASKAR M K, CHATTERJI B N. Advanced wavelet transform for image processing-a survey[M]//MANDAL J, BHATTACHARYA K, MAJUMDAR I, et al, eds. Information, photonics and communication. Lecture Notes in Networks and Systems. Singapore: Springer, 2019.
28	WANG Z, BOVIK A C, SHEIKH H R , et al. Image quality assessment: from error visibility to structural similarity[J]. IEEE Trans. on Image Processing, 2004, 13: 600−612.
29	SAMAJDAR T, QURAISHI M I. Analysis and evaluation of image quality metrics[M]//MANDAL J, SATAPATHY S, KUMAR SANYAL M, et al. eds. Information systems design and intelligent applications. Advances in Intelligent Systems and Computing. New Delhi: Springer India, 2015.
30	LIU A, LIN W, NARWARIA M. Image quality assessment based on gradient similarity[J]. IEEE Trans. on Image Processing, 2012, 21, 1500- 1512. doi: 10.1109/TIP.2011.2175935

极化方式	海情	RMSE	PSNR	与原图均值差值均值	与原图方差差值均值	径向积分相似度均值
HH	3级	30.79122365	18.84552259	3.728261104	69.24934308	0.994451753
	4级	36.1006545	17.55053249	4.71346784	95.53434512	0.994136654
	5级	42.48723	15.82286	6.48229	110.75557	0.99363
HV	3级	33.38296543	17.99349767	8.429126783	334.0068638	0.987269033
	4级	40.42520566	16.74161314	12.21861197	630.1909487	0.98747514
	5级	46.09461	15.38488	14.59328	698.26031	0.9856
VH	3级	34.24186047	17.81118238	7.400956702	300.4465602	0.988056022
	4级	40.56792253	16.60335548	11.88713918	525.8215773	0.988192898
	5级	41.51968	16.21547	10.72215	562.63077	0.98694
VV	3级	27.26850499	19.83796746	2.968774892	51.59439813	0.994560809
	4级	31.94759591	18.61335832	3.566741086	85.94822673	0.99434258
	5级	38.15552	16.77791	5.21796	110.54041	0.99404

[1]	花思齐, 赵伟, 刘建业. 基于改进滤波器和图像多尺度变换的背景抑制算法[J]. 系统工程与电子技术, 2020, 42(8): 1679-1684.
[2]	薛源, 杨梓鑫, 徐浩军, 王国智. 基于网格吸附插值法的不规则尾流场构建方法[J]. 系统工程与电子技术, 2019, 41(4): 870-875.

基于方差拟合插值的SAR图像数据集扩充方法

Variance-fitting interpolation-based method for SAR image dataset expansion

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 16

参考文献 30

相关文章 2

编辑推荐

Metrics

本文评价