[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-2176":3},{"id":4,"title":5,"url":6,"summary":7,"summary_zh":8,"content":9,"source_name":10,"source_url":6,"published_at":11,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":15,"score_detail":16,"sources":24,"tags":26,"view_count":32,"doi":33,"paper":34,"created_at":54},2176,"A Robust and Efficient Scheme for Real-Time SAR Speckle Noise Reduction based on Attentional Self-Supervised Contrastive Deep Learning","https:\u002F\u002Fdoi.org\u002F10.17725\u002Fj.rensit.2026.18.429","Synthetic aperture radar (SAR) is a critical and essential technology for remote sensing, achieving high resolution under all weather and day\u002Fnight conditions. But SAR images suffer from multiplicative speckle noise, making the images poor in quality and resulting in adverse effects on the image interpretation and automatic analysis. In order to solve this problem, a novel Residual Attention Swin Transformer Network (RAST-Net) is proposed to efficiently despeckle the SAR images.To improve local feature representation, the suggested architecture combines residual convolutional learning with the Convolutional Block Attention Module (CBAM); light Swin Transformer blocks are added to catch long-range contextual dependencies.Additionally, the self-supervised contrastive learning method is applied to enhance the capacity to differentiate features and achieve generalization without much dependence on clear reference data. The techniques of quantization-aware optimization are also incorporated into the proposed design to ensure the successful implementation of the method on real devices like NVIDIA Jetson Orin NX, and Xilinx ZynqUltraScale+ MPSoC FPGA systems. Comprehensive experimental results indicate that the proposed model demonstrates better performance than conventional filters, CNN networks, self-supervised models, and some other recently developed transformers and diffusion-based approaches operating with various objective indices such as PSNR, SSIM, FSIM ENL, and EPI.The proposed network not only achieves high restoration accuracy, but also has moderate computational complexity, low inference delay and excellent deployment efficiency on embedded hardware platform. Based on these results, the proposed RAST-Net is an effective and practical solution for real-time SAR image restoration in intelligent remote sensing and edge AI applications.","合成孔径雷达（SAR）是一种关键且必不可少的遥感技术，能够在全天候、全天时条件下实现高分辨率成像。但SAR图像受到乘性斑点噪声的影响，导致图像质量下降，并对图像解译和自动分析产生不利影响。为了解决这一问题，提出了一种新型残差注意力Swin Transformer网络（RAST-Net），用于高效地对SAR图像进行去斑。为提升局部特征表达能力，所提出的架构将残差卷积学习与卷积块注意力模块（CBAM）相结合；并加入轻量Swin Transformer块以捕获长距离上下文依赖关系。此外，采用自监督对比学习方法增强特征区分能力，并在不过度依赖清晰参考数据的情况下实现泛化。该设计还融入了量化感知优化技术，以确保该方法能够在NVIDIA Jetson Orin NX和Xilinx ZynqUltraScale+ MPSoC FPGA系统等真实设备上成功部署。综合实验结果表明，在PSNR、SSIM、FSIM、ENL和EPI等多种客观指标下，所提出的模型相比传统滤波器、CNN网络、自监督模型以及一些其他近期开发的Transformer和基于扩散的方法表现出更好的性能。所提出的网络不仅实现了较高的恢复精度，而且具有适中的计算复杂度、较低的推理延迟以及在嵌入式硬件平台上的出色部署效率。基于这些结果，所提出的RAST-Net是智能遥感和边缘AI应用中实时SAR图像恢复的一种有效且实用的解决方案。",null,"Radioelectronics Nanosystems Information Technologies","2026-09-08T00:00:00Z","论文",10,false,75,{"impact":17,"substance":18,"depth":19,"authority":20,"freshness":21,"relevant":22,"comment":23},16,21,18,12,8,1,"提出RAST-Net自监督对比学习去斑方案并在嵌入式平台验证，方法新颖、指标充分，对农业遥感影像智能解译有实用价值，但属细分技术进展，影响层级有限。",[25],{"name":10,"url":6},[27,28,29,30,31],"智慧农业","农业人工智能","边缘计算","遥感","SAR影像",0,"10.17725\u002Fj.rensit.2026.18.429",{"doi":33,"openalex_id":35,"authors":36,"venue":10,"cited_by_count":32,"oa_url":46,"card":47,"direction":51,"ingested_from":53},"W7211960754",[37,39,42,44],{"name":38,"orcid":9},"Marwah M. Hassooni",{"name":40,"orcid":41},"Abdullah Al-Obaidi","https:\u002F\u002Forcid.org\u002F0000-0002-6541-8999",{"name":43,"orcid":9},"Mustafa Nadhim Ghazal",{"name":45,"orcid":9},"Anas Fouad Ahmed.","http:\u002F\u002Fen.rensit.ru\u002Fvypuski\u002Farticle\u002F774\u002F18(4)429-442e.pdf",{"tldr":48,"method":49,"finding":50,"direction":51,"opportunity":52},"提出RAST-Net，用注意力自监督对比深度学习实现SAR图像实时去斑。","残差卷积+CBAM+轻量Swin Transformer，自监督对比学习与量化优","在多项指标上优于传统滤波、CNN、自监督及扩散方法，且嵌入式部署高效。","农业遥感与作物表型","可探索该轻量去斑模型在农业SAR作物监测边缘设备上的迁移与实时应用。","openalex","2026-09-11T23:30:34.257413Z"]