CFM-SAR: A Class-Conditioned Flow Matching with SCM-UNet for SAR Image Generation
DOI:
https://doi.org/10.4108/eetinis.133.14079Keywords:
Class-based prior, flow matching, SAR image classification, SAR image generation, SCM-UNetAbstract
Monitoring and analyzing sea-ice changes is critically important for navigation through polar regions, ecosystem monitoring, and environmental risk assessment, and Synthetic Aperture Radar(SAR) imagery is one of the most reliable and useful data sources for this goal. SAR images exhibit complex backscatter patterns that are generally not interpretable without technical expertise. Consequently, accurate annotation of SAR images requires domain experts to identify the corresponding sea ice classes, resulting in labelled datasets that are both limited and often imbalanced across classes.A further challenge is that SAR image channels differ from those of conventional images, with HH, HV, and incidence angle each containing important physical information. Although current image generation methodologies have strong capabilities in RGB images and have solved several limitations, including training instability and high computational cost, some challenges in SAR image generation remain. These issues, such as limited preservation of channel-specific physical information and mode collapse, motivate us to work in generative modelling and seek to overcome them. In this study, we propose a Flow Matching framework with a class-informed prior. Instead of using a standard Gaussian distribution as the initial point for learning, the Class-Conditioned Flow Matching (CFM) model estimates the mean and variance of each image class and uses them to better approximate the data distribution from a more informative starting point. Another contribution of this paper is the proposed Swin Transformer Convolutional Hybrid (SCM-UNet) model, which generates images using a modified U-Net architecture. Experiments on the chosen dataset demonstrate that the proposed model improves visual metrics in generated images (FSD=19.23, PSNR=28.0, and SSIM=0.83) and effectively enhances the downstream task. Results show that, when trained exclusively on generated data and evaluated on real SAR images, the classification model achieved an average accuracy of 91.61% and an F1-score exceeding 91.5%.
Downloads
References
[1] Rodríguez JP, Klemm K, Duarte CM, Eguíluz VM. Shipping traffic through the Arctic Ocean: Spatial distribution, seasonal variation, and its dependence on the sea ice extent. iScience. 2024 July;27:110236.
[2] Zakhvatkina N, Smirnov V, Bychkova I. Satellite SAR Data-Based Sea Ice Classification: An Overview. Geosciences. 2019;9(4):152.
[3] Choi H, Jeong J. Speckle Noise Reduction Technique for SAR Images Using Statistical Characteristics of Speckle Noise and Discrete Wavelet Transform. Remote Sensing. 2019;11(10):1184.
[4] Soumekh M. Synthetic Aperture Radar Signal Processing with MATLAB Algorithms. New York, USA: Wiley-Interscience; 1999.
[5] Peter E, Ang LM, Seng KP, Srivastava S. Recent Advances in Deep Learning for SAR Images: Overview of Methods, Challenges, and Future Directions. Sensors. 2026;26(4):1143. Available from: https://doi.org/10.3390/s26041143.
[6] Boulze H, Korosov A, Brajard J. Classification of Sea Ice Types in Sentinel-1 SAR Data Using Convolutional Neural Networks. Remote Sensing. 2020;12(13):2165. Available from: https://doi.org/10.3390/rs12132165.
[7] Maillard P, Clausi DA, Deng H. Operational Map-Guided Classification of SAR Sea Ice Imagery. IEEE Transactions on Geoscience and Remote Sensing. 2005 Dec;43(12):2940-51.
[8] Radhakrishnan K, Scott KA, Clausi DA. Sea Ice Concentration Estimation: Using Passive Microwave and SAR Data With a U-Net and Curriculum Learning. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. 2021;14:5339-51.
[9] Ricker R, Hendricks S, Kaleschke L, Tian-Kunze X, King J, Haas C. A Weekly Arctic Sea-Ice Thickness Data Record from Merged CryoSat-2 and SMOS Satellite Data. The Cryosphere. 2017;11(4):1607-23.
[10] Komarov AS, Barber DG. Sea Ice Motion Tracking From Sequential Dual-Polarization RADARSAT-2 Images. IEEE Transactions on Geoscience and Remote Sensing. 2014;52(7):4431-41.
[11] Ressel R, Frost A, Lehner S. A Neural Network- Based Classification for Sea Ice and Open Water Detection Using Sentinel-1 SAR Data. Remote Sensing. 2016;8(12):1006.
[12] Power D, Youden J, Lane K, Randell C, Flett D. Iceberg Detection Capabilities of RADARSAT Synthetic Aperture Radar. Canadian Journal of Remote Sensing. 2001;27(5):476-86.
[13] Gascard JC, Riemann-Campe K, Gerdes R, Schyberg H, Randriamampianina R, Karcher M, et al. Future sea ice conditions and weather forecasts in the Arctic: Implications for Arctic shipping. Ambio. 2017;46(Suppl 3):355-67.
[14] Løset S, Shkhinek K, Gudmestad OT, Strass P, Michalenko E, Frederking R, et al. Comparison of the Physical Environment of Some Arctic Seas. Cold Regions Science and Technology. 1999;29(3):201-14.
[15] Kwok R. Arctic Sea Ice Thickness, Volume, and Multiyear Ice Coverage: Losses and Coupled Variability (1958–2018). Environmental Research Letters. 2018;13(10):105005.
[16] Aase J. The Ortelius Incident in the Hinlopen Strait— A Case Study on How Satellite-Based AIS Can Support Search and Rescue Operations in Remote Waters. Resources. 2017;6(3):35.
[17] Passah A, Sur SN, Paul B, Kandar D. SAR Image Classification: A Comprehensive Study and Analysis. IEEE Access. 2022;10:20385-99.
[18] Khaleghian S, Ullah H, Kræmer T, Hughes N, Eltoft T, Marinoni A. Sea Ice Classification of SAR Imagery Based on Convolution Neural Networks. Remote Sensing. 2021 Apr;13(9):1734.
[19] Goodfellow I, Pouget-Abadie J, Mirza M, Xu B, Warde- Farley D, Ozair S, et al. Generative adversarial nets. In: Advances in Neural Information Processing Systems; 2014. p. 2672-80.
[20] Ho J, Jain A, Abbeel P. Denoising diffusion probabilistic models. In: Advances in Neural Information Processing Systems. vol. 33; 2020. p. 6840-51.
[21] Lipman Y, Chen RTQ, Ben-Hamu H, Nickel M, Le M. Flow matching for generative modeling. In: International Conference on Learning Representations (ICLR); 2023. .
[22] Esser P, Rombach R, Ommer B. Taming Transformers for High-Resolution Image Synthesis. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR); 2021. p. 12873-83.
[23] Wang Z, Zhang Z, Shan X, Wei H, Tang P. Generative models for SAR–optical image translation: A systematic review. International Journal of Applied Earth Observation and Geoinformation. 2026 February;146:105009.
[24] Passah A, Sur SN, Abraham A, Kandar D. Synthetic aperture radar image analysis based on deep learning: A review of a decade of research. Engineering Applications of Artificial Intelligence. 2023;123:106305.
[25] Stofa MM, Abdani SR, Moubark AM, Zainuri MAAM, Ibrahim AA, Kamari NAM, et al. Recent developments of artificial intelligence methods for sea ice concentration monitoring using high-resolution imaging datasets. Ecological Informatics. 2025;87:103132.
[26] Wan H, Luo X, Wu Z, Qin X, Chen X, Li B, et al. Multi-featured sea ice classification with SAR image based on convolutional neural network. Remote Sensing. 2023;15(16):4014.
[27] Stokholm A, Wulf T, Kucik A, Saldo R, Buus- Hinkler J, Hvidegaard SM. AI4SeaIce: Toward Solving Ambiguous SAR Textures in Convolutional Neural Networks for Automatic Sea Ice Concentration Charting. IEEE Transactions on Geoscience and Remote Sensing. 2022;60:1-13.
[28] Song Q, Xu F, Jin YQ. SAR Image Representation Learning With Adversarial Autoencoder Networks. In: 2019 IEEE International Geoscience and Remote Sensing Symposium (IGARSS). Yokohama, Japan: IEEE; 2019. p. 9498-501.
[29] Liu K, Qiu G, Tang W, Zhou F. Spectral regularization for combating mode collapse in GANs. Image and Vision Computing. 2020;104:104019.
[30] Oghim S, Kim Y, Bang H, Lim D, Ko J. SAR image generation method using DH-GAN for automatic target recognition. Sensors. 2024;24(2):670.
[31] Chen Y, Zhu Z, Huang Y, Wang P, Huang B, Mura MD. MSF: A Multi-Scale Fusion Generative Adversarial Network for SAR-to-Optical Image Translation. In: IGARSS 2024 - 2024 IEEE International Geoscience and Remote Sensing Symposium; 2024. p. 9058-61.
[32] Hwang J, Yu C, Shin Y. SAR-to-Optical Image Translation Using SSIM and Perceptual Loss Based Cycle-Consistent GAN. In: 2020 International Conference on Information and Communication Technology Convergence (ICTC); 2020. p. 191-4.
[33] Qosja D, Wagner S, O’Hagan D. SAR image synthesis with diffusion models. In: Proc. IEEE Radar Conference (RadarConf); 2024. p. 1-6.
[34] Wang J, Yang H, Liu Z, Chen H. SSDDPM: A single SAR image generation method based on denoising diffusion probabilistic model. Scientific Reports. 2025;15(10867).
[35] Tong A, Fatras K, Malkin N, Huguet G, Zhang Y, Rector- Brooks J, et al. Improving and generalizing flow-based generative models with minibatch optimal transport. Transactions on Machine Learning Research. 2024.
[36] Ben-Hamu H, Puny O, Gat I, Karrer B, Singer U, Lipman Y. D-Flow: Differentiating through Flows for Controlled Generation. In: Proceedings of the 41st International Conference on Machine Learning. vol. 235. PMLR; 2024. p. 3462-83.
[37] Kong L, Tao M, Liu Y, Wang B, Fu J, Wang CC, et al. AlignFlow: Improving Flow-Based Generative Models with Semi-Discrete Optimal Transport. In: Proceedings of the 14th International Conference on Learning Representations (ICLR). Rio de Janeiro, Brazil; 2026.
[38] Jin Y, Sun Z, Li N, Xu K, Xu K, Jiang H, et al. Pyramidal Flow Matching for Efficient Video Generative Modeling. In: Proceedings of the 13th International Conference on Learning Representations (ICLR). Singapore; 2025. .
[39] Peebles W, Xie S. Scalable Diffusion Models with Transformers. In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). Paris, France; 2023. p. 4195-205.
[40] Beizaee F, Lodygensky GA, Adamson CL, Thompson DK, Cheong JLY, Spittle AJ, et al. Harmonizing flows: Leveraging normalizing flows for unsupervised and source-free MRI harmonization. Medical Image Analysis. 2025;101:103483.
[41] Kobyzev I, Prince SJD, Brubaker MA. Normalizing Flows: An Introduction and Review of Current Methods. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2021;43(11):3964-79.
[42] Feng R, Yu C, Deng W, Hu P, Wu T. On the Guidance of Flow Matching. In: Proceedings of the 42nd International Conference on Machine Learning (ICML). Vancouver, BC, Canada; 2025. .
[43] Baramaki N, Le QN, McNiven BDE, Fahim M. SCMNet: A Lightweight AI-Based Sea Ice Classification for Climate Change. EAI Endorsed Transactions on Industrial Networks and Intelligent Systems. 2026 Jun;13(2). Available from: https://doi.org/10.4108/eetinis.132.12246.
[44] Heusel M, Ramsauer H, Unterthiner T, Nessler B, Hochreiter S. GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium. In: Advances in Neural Information Processing Systems (NeurIPS). vol. 30; 2017. .
[45] Betzalel E, Penso C, Fetaya E. Evaluation Metrics for Generative Models: An Empirical Study. Machine Learning and Knowledge Extraction. 2024;6(3):1531-44.
[46] Kynkäänniemi T, Karras T, Aittala M, Aila T, Lehtinen J. The Role of ImageNet Classes in Fréchet Inception Distance. In: International Conference on Learning Representations (ICLR); 2023.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Nazanin Baramaki, Quang Nhat Le, Muhammad Fahim, Tuan Thanh Nguyen
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
This is an open-access article distributed under the terms of the Creative Commons Attribution CC BY 3.0 license, which permits unlimited use, distribution, and reproduction in any medium so long as the original work is properly cited.
