基于改进Transformer的综合孔径辐射计重构算法

doi:10.16180/j.cnki.issn1007-7820.2025.08.013

电子科技 ›› 2025, Vol. 38 ›› Issue (8): 94-100.doi: 10.16180/j.cnki.issn1007-7820.2025.08.013

• • 上一篇

基于改进Transformer的综合孔径辐射计重构算法

程伟豪¹^,², 杨晓城¹^,²(), 武林²^,³, 阎敬业²^,³, 蒋明峰¹, 魏波¹

1.浙江理工大学计算机科学与技术学院,浙江杭州 310018
2.中国科学院空间天气学国家重点实验室,北京 100190
3.中国科学院国家空间科学中心,北京 100190

收稿日期:2025-03-13 修回日期:2025-04-14 出版日期:2025-08-15 发布日期:2025-07-10
通讯作者: 杨晓城(1988-),男,E-mail:yangxiaoch209@163.com,博士,讲师。研究方向:干涉与综合孔径技术、图像处理、天文技术与方法。
作者简介:程伟豪(1999-),男,硕士研究生。研究方向:综合孔径辐射计图像重构。
基金资助:
国家重点实验室专项基金(202528);国家自然科学基金(62272415);国家自然科学基金(62101497)

Reconstruction Algorithm of Synthetic Aperture Interferometer Radiometer Based on Improved Transformer

CHENG Weihao¹^,², YANG Xiaocheng¹^,²(), WU Lin²^,³, YAN Jingye²^,³, JIANG Mingfeng¹, WEI Bo¹

1. School of Computer Science and Technology,Zhejiang Sci-Tech University,Hangzhou 310018,China
2. State Key Laboratory of Space Weather,Chinese Academy of Sciences,Beijing 100190,China
3. National Space Science Center,Chinese Academy of Sciences,Beijing 100190,China

Received:2025-03-13 Revised:2025-04-14 Online:2025-08-15 Published:2025-07-10
Supported by:
Specialized Research Fund for State Key Laboratories(202528);National Natural Science Foundation of China(62272415);National Natural Science Foundation of China(62101497)

摘要/Abstract

摘要：

在综合孔径微波辐射计(Synthetic Aperture Interferometer Radiometer, SAIR)中,利用测量可见度数据重构观测图像是一个不适定逆问题。针对目前图像重构方法存在较大残余误差和振荡伪影问题,文中提出了一种基于改进Transformer的SAIR重构方法。通过预处理模块从可见度函数提取浅层特征,再经深层特征提取模块提取可见度函数的深层特征,由SAIR图像重构模块得到结果。与传统Transformer结构相比,所提改进Transformer方法采用U-Net结构,充分利用可见度函数的多尺度信息进行图像重构,同时从通道维度对特征应用自注意力机制,减少了信息丢失。实验结果表明,所提方法在重构质量和噪声抑制方面优于传统正则化方法和深度学习方法,为SAIR图像重构提供了一种有效的解决方案。

关键词: 辐射计, 综合孔径, 逆问题, 图像重构, 深度学习, Transformer, 注意力机制, 噪声抑制

Abstract:

In SAIR(Synthetic Aperture Imaging Radiometer), reconstructing the observed image from the measured visibility data is an ill-posed inverse problem. Aiming at the problems of large residual errors and oscillatory artifacts existing in the current image reconstruction methods, this study proposes a SAIR reconstruction method based on the improved Transformer. Shallow features are extracted from the visibility function through the preprocessing module, and then the deep features of the visibility function are extracted by the deep feature extraction module, and the result is obtained by the SAIR image reconstruction module. Compared with the traditional Transformer structure, the proposed improved Transformer method adopts the U-Net structure, which makes full use of the multi-scale information of the visibility function for image reconstruction. At the same time, the self-attention mechanism is applied to the features from the channel dimension, reducing information loss. The experimental results show that the proposed method outperforms the traditional regularization methods and deep learning methods in terms of reconstruction quality and noise suppression, providing an effective solution for SAIR image reconstruction.

Key words: radiometer, synthetic aperture, inverse problem, image reconstruction, deep learning, Transformer, attention mechanism, noise suppression

中图分类号:

TP722.6

程伟豪, 杨晓城, 武林, 阎敬业, 蒋明峰, 魏波. 基于改进Transformer的综合孔径辐射计重构算法[J]. 电子科技, 2025, 38(8): 94-100.

CHENG Weihao, YANG Xiaocheng, WU Lin, YAN Jingye, JIANG Mingfeng, WEI Bo. Reconstruction Algorithm of Synthetic Aperture Interferometer Radiometer Based on Improved Transformer[J]. Electronic Science and Technology, 2025, 38(8): 94-100.

图/表 10

图1

图2

图3

图4

图5

表1

图6

表2

表3

图7

参考文献 22

[1]	Le Vine D M, Griffis A J, Swift C T, et al. ESTAR:A synthetic aperture microwave radiometer for remote sensing applications[J]. Proceedings of the IEEE, 1994, 82(12):1787-1801.
[2]	Tanner A B, Wilson W J, Lambrigsten B H, et al. Initial results of the geostationary synthetic thinned array radiometer demonstrator instrument[J]. IEEE Transactions on Geoscience and Remote Sensing, 2007, 45(7):1947-1957.
[3]	Zhang C, Liu H, Wu J, et al. Imaging analysis and first results of the geostationary interferometric microwave sounder demonstrator[J]. IEEE Transactions on Geoscience and Remote Sensing, 2014, 53(1):207-218.
[4]	Corbella I, Duffo N, Vall-Lossera M, et al. The visibility function in interferometric aperture synthesis radiometry[J]. IEEE Transactions on Geoscience and Remote Sensing, 2004, 42(8):1677-1682.
[5]	Anterrieu E. A resolving matrix approach for synthetic aperture imaging radiometers[J]. IEEE Transactions on Geoscience and Remote Sensing, 2004, 42(8):1649-1656.
[6]	杨真乙, 杨晓城, 阎敬业, 等. 基于SP-LSQR的综合孔径辐射成像方法[J]. 微波学报, 2023, 39(2):16-20.
	Yang Zhenyi, Yang Xiaocheng, Yan Jingye, et al. Imaging method of synthetic aperture interferometric radiometers based on subspace preconditioned LSQR[J]. Journal of Microwaves, 2023, 39(2):16-20.
[7]	Yang X, Lu C, Yan J, et al. Reweighted total variation regularization based on split Bregman in synthetic aperture imaging radiometry[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19(2):1-5.
[8]	杨晓城, 杨真乙, 阎敬业, 等. 基于拉普拉斯算子的综合孔径辐射计成像算法[J]. 电子学报, 2022, 50(2):339-345. doi: 10.12263/DZXB.20201283
	Yang Xiaocheng, Yang Zhenyi, Yan Jingye, et al. Imaging algorithm of synthetic aperture interferometric radiometers based on Laplace operator[J]. Acta Electronica Sinica, 2022, 50(2):339-345. doi: 10.12263/DZXB.20201283
[9]	Wu L, Hu F, He F, et al. Bayesian inference for inversion in synthetic aperture imaging radiometry[J]. IEEE Geoscience and Remote Sensing Letters, 2016, 13(8):1049-1053.
[10]	Zhu T, Yang X, Yan J, et al. A map approach with Huber-MRF for synthetic aperture interferometric radiometer[J]. IEEE Geoscience and Remote Sensing Letters, 2024, 21(10): 1-5.
[11]	Zhang Y, Ren Y, Miao W, et al. Microwave SAIR imaging approach based on deep convolutional neural network[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(12): 10376-10389.
[12]	Xiao C, Dou H, Li H, et al. Image reconstruction of synthetic aperture radiometer by transformer[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61(2):1-15.
[13]	Young T, Hazarika D, Poria S, et al. Recent trends in deep learning based natural language processing[J]. IEEE Computational Intelligence Magazine, 2018, 13(3):55-75.
[14]	陈谢发, 李敏, 赵晶钰, 等. 基于双层注意力机制的遥感小目标检测算法[J]. 火箭军工程大学学报, 2025, 39(1):60-66.
	Chen Xiefa, Li Min, Zhao Jingyu, et al. Remote sensing small object detection algorithm based on dual-layer attention mechanism[J]. Journal of Rocket Force University of Engineering, 2025, 39(1):60-66.
[15]	朱灵龙, 王亚刚, 陈怡. 融合Transformer与卷积神经网络的图像分类算法[EB/OL].(2025-02-24)[2025-03-13]. https://doi.org/10.16180/j.cnki.issn1007-7820.2025.10.012.
	Zhu Linglong, Wang Yagang, Chen Yi. Research on image classification algorithm combining Transformer with CNN[EB/OL].(2025-02-24)[2025-03-13]. https://doi.org/10.16180/j.cnki.issn1007-7820.2025.10.012.
[16]	Liu Y, Zhang Y, Wang Y, et al. A survey of visual transformers[J]. IEEE Transactions on Neural Networks and Learning Systems, 2023, 35(6):7578-7498.
[17]	陈宇洋, 李峰. 融合CNN与Transformer的视网膜OCT图像积液分割方法[J]. 电子科技, 2025, 38(3):47-59.
	Chen Yuyang, Li Feng. Integration of CNN and Transformer for retinal OCT image fluid segmentation method[J]. Electronic Science and Technology, 2025, 38(3):47-59.
[18]	Bao G, Graeber M B, Wang X. Depthwise multiception convolution for reducing network parameters without sacrificing accuracy[C]. Shenzhen: The Sixteenth International Conference on Control,Automation, Robotics and Vision, 2020:747-752.
[19]	张慧, 李菲菲. 基于多模态交叉互动的情感识别算法[J]. 电子科技, 2024, 37(10):81-87.
	Zhang Hui, Li Feifei. Emotion recognition algorithm based on multimodal cross-Interaction[J]. Electronic Science and Technology, 2024, 37(10):81-87.
[20]	李智, 高连如, 郑珂, 等. 高分辨率遥感图像场景分类研究进展[J]. 遥感学报, 2024, 28(11):2739-2760.
	Li Zhi, Gao Lianru, Zheng Ke, et al. Research progress of high-resolution remote sensing image scene classification[J]. National Remote Sensing Bulletin, 2024, 28(11):2739-2760.
[21]	Yang X, Lu C, Yan J, et al. Adaptive LP reconstruction method for synthetic aperture interferometric radiometer[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2022, 15(5):3282-3291.
[22]	Yang X, Lu C, Yan J, et al. Iterative regularization using an acceleration technique for synthetic aperture interferometric radiometer[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2023, 16(7):4649-4657.

参数	数值
中心频率	f₀=1.4 GHz
天线阵列	均匀T型
天线数量	46
带宽	B=20 MHz
最小基线	Δd=0.6λ
积分时间接收机噪声温度	τ=1 s
接收机噪声温度	T_R=100 K
天线温度	T_A=250 K

方法	建筑物			棒球场
方法	RMSE/K	PSNR/dB	RMSE/K		PSNR/dB
带限正则化	16.59	24.21	14.48	24.96
SAIR-CNN	18.39	23.31	16.40	23.89
ASR-T	16.45	24.28	14.14	25.17
本文方法	14.43	25.42	11.76	26.77

方法	评价指标
方法	RMSE/K	PSNR/dB
带限正则化	15.34	23.65
SAIR-CNN	15.95	23.20
ASR-T	10.29	25.49
本文方法	9.44	26.48

基于改进Transformer的综合孔径辐射计重构算法

Reconstruction Algorithm of Synthetic Aperture Interferometer Radiometer Based on Improved Transformer

RichHTML

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 22

相关文章 15

Metrics

本文评价

推荐阅读 0

[1]	熊子恒, 张轩雄. 基于深度学习的红外与可见光图像匹配[J]. 电子科技, 2025, 38(9): 1-8.
[2]	龚宇, 吴鹏. 基于交叉融合编码器的Transformer图像特征提取网络[J]. 电子科技, 2025, 38(9): 20-25.
[3]	齐浩东, 程晓颖, 李海生, 陈晓, 徐鑫炯. 基于改进YOLOv7的焊缝缺陷检测算法[J]. 电子科技, 2025, 38(9): 79-84.
[4]	魏擎擎, 鲁玉军. 基于改进Faster R-CNN的铝材表面缺陷检测[J]. 电子科技, 2025, 38(9): 85-92.
[5]	赵开, 胡春燕, 李菲菲. 基于高分辨率网络的轻量化人体姿态估计算法[J]. 电子科技, 2025, 38(9): 93-100.
[6]	陈清江, 李宗莹. 结合并联Transformer和残差U-Net网络的水下图像增强[J]. 电子科技, 2025, 38(8): 57-65.
[7]	沈依宁, 王一然, 吴聪. 基于深度学习的关系抽取研究进展[J]. 电子科技, 2025, 38(7): 40-49.
[8]	唐玉良, 张轩雄. 基于深度学习的车位检测方[J]. 电子科技, 2025, 38(6): 23-29.
[9]	陈清江, 杨双. 基于多层特征增强的双分支图像去雾算法[J]. 电子科技, 2025, 38(6): 30-38.
[10]	刘雪莉, 杜洪波, 袁雪丰, 朱立军. 基于CBAM残差块结合纹理采样器的水墨迁移算法[J]. 电子科技, 2025, 38(6): 65-73.
[11]	陈漫漫, 于莲芝. 结合共注意网络的深度BiGRU和DPCS情感分析模型[J]. 电子科技, 2025, 38(5): 22-30.
[12]	胡锦磊, 田恩刚, 瞿枫. 融合通道全局注意力机制的双路径实时语义分割[J]. 电子科技, 2025, 38(5): 83-88.
[13]	赵文琪, 张立新, 阚希, 郑好韧. 基于双分支网络的户外垃圾检测识别[J]. 电子科技, 2025, 38(4): 1-9.
[14]	毕含嘉, 杨楚睿, 王小雨, 黄悦华. 基于改进YOLOv7的输电线路多类缺陷目标检测[J]. 电子科技, 2025, 38(4): 16-24.
[15]	傅鹏, 宋晓霞. 基于CycleGAN的地震数据去噪方法[J]. 电子科技, 2025, 38(4): 25-30.