The results support the conclusion that a lightweight 3D geometric prior improves viewpoint adherence for controllable SAR generation; it is intended as generation guidance rather than high-fidelity electromagnetic construction.
Abstract
Synthetic aperture radar (SAR) image generation can mitigate data scarcity, but controllablegeneration under sparse observation angles remains difficult. Recent SAR generative studies im-prove texture realism, yet explicit geometry-aware control is still limited. This paper studiesthe focused and verifiable setting of intermediate-azimuth completion: 3D-model-derived geo-metric priors guide a diffusion model to synthesize the views missing from sparse-angle trainingdata. GeoDiff-SAR constructs a lightweight multi-bounce ray-tracing prior, encodes the result-ing point cloud, and fuses it with text conditioning while adapting Stable Diffusion 3.5 Mediumthrough low-rank adaptation. On a real four-category aircraft dataset, GeoDiff-SAR reaches anSSIM of 0.812 and azimuth consistency of 0.940, compared with 0.738 and 0.782 for the text-conditioned SD3.5 Medium baseline. The same sparse-angle protocol on five MSTAR vehicleclasses yields an SSIM of 0.878 and azimuth consistency of 0.917. These results support theconclusion that a lightweight 3D geometric prior improves viewpoint adherence for controllableSAR generation; it is intended as generation guidance rather than high-fidelity electromagneticreconstruction.
Polarimetric Synthetic Aperture Radar (PolSAR) classification underpins all-weather Earth observation. Conventional Wishart methods depend on rigid handcrafted operators with limited adaptability, while mainstream deep networks ignore PolSAR native Wishart scattering statistics. Additionally, fixed convolution windows fail to capture multi-scale, multi-directional terrain patterns, harming boundary detection and small-object characterization. To mitigate these drawbacks, we propose WGDNet, a Wishart-guided geometric-aware deep network. It integrates three core designs: (1) learnable Wishart convolutions with directional kernels for multi-scale statistical edge feature extraction; (2) an orientation-prior aggregation module that estimates dominant local directions and confidences to refine directional Wishart outputs adaptively; (3) GAnet, a scale-direction adaptive geometric-aware convolution that dynamically reshapes sampling grids to model anisotropic terrain and retain fine details. Our contributions lie in learnable Wishart statistical modeling, orientation-prior feature aggregation, and geometry-adaptive convolution. Evaluations across four real PolSAR datasets verify WGDNet surpasses existing state-of-the-art approaches in classification accuracy and boundary fidelity.
ABSTRACT Synthetic aperture radar (SAR) aircraft classification remains challenging, as measured SAR images are affected by imaging conditions and speckle noise, while the task is further complicated by class imbalance and limited labelled data. Simulation offers a potential route of mitigating these challenges; however, the physically meaningful integration of such priors into SAR aircraft classification remains an open problem. To address this issue, this paper proposes a retrieval-assisted feature fusion framework for SAR aircraft classification, termed RAFF. The framework first constructs Bounce-Coded Radar Cross-Section (BCRCS) images from electromagnetic simulation, in which geometry-related scattering priors are encoded. A Pre-Classification module then generates candidate classes and semantic guidance for each query SAR image. On this basis, a Perceptual-Similarity Retrieval module establishes class-constrained correspondences between measured SAR images and simulated priors using Learned Perceptual Image Patch Similarity (LPIPS), without requiring strict pixel-level alignment. Finally, PhysCrossNet adaptively fuses measured SAR features, retrieved BCRCS priors, and semantic cues for final classification. Experimental results indicate that RAFF attains the strongest performance on the imbalanced SAR-Aircraft-1.0 dataset and remains highly competitive on the more balanced SAR-ACD dataset. The framework also demonstrates enhanced robustness under progressively reduced training data. Supplementary analyses further reveal that BCRCS provides the principal discriminative gain, semantic guidance improves fusion stability, and the decision regions of RAFF align with physically meaningful scattering structures. These results confirm the effectiveness of RAFF and support the use of retrieval-assisted fusion of simulation-derived physical priors for SAR aircraft classification.
Shangchen Feng, Xikai Fu, Yan-Lin Feng et al.· International Journal of Rem...· 0 citations
Synthetic aperture radar (SAR) ship detection is often limited by the quantity and distribution coverage of labeled training data. Under this condition, SAR ship image generation for augmentation should not be treated as a purely visual synthesis problem. For downstream detector training, the generated samples should preserve controllable target layout, SAR-domain compatibility, and sufficient consistency with real SAR observations. Existing methods, however, still have difficulty satisfying these requirements jointly, which limits the usefulness of synthetic samples for SAR ship detection. To address this issue, this article proposes STARS-SAR, a structure-aware diffusion framework for detector-oriented SAR ship augmentation. Instead of treating structural controllability and SAR-domain adaptation as separate objectives, STARS-SAR integrates them into a unified generation process through structure-guided cross-attention modulation, an SAR-oriented LoRA adaptation strategy, and region-aware SAR texture alignment. In this way, the generated samples remain more compatible with real SAR observations while preserving local target–background coherence. Experiments on HRSID and SSDD show that STARS-SAR generates representative SAR ship images with improved structural fidelity and better compatibility with real SAR observations. Under limited-data training settings, the generated samples also serve as effective augmentation data and improve downstream ship detection in both in-domain and cross-domain settings. These results show that STARS-SAR is effective for detector-oriented SAR ship image augmentation under practical data-limited conditions.
Pingpeng Tang, Xiangyu Zhang, Qiao Shi et al.· IEEE Journal of Selected Top...· 0 citations
Protecting target regions from reconnaissance is a critical task in synthetic aperture radar (SAR) countermeasures. For large facilities such as airports and key infrastructure, deceptive scene jamming requires more than inserting isolated false targets. It requires a credible scene template that can project the protected region as a coherent false SAR scene compatible with the surrounding background. In repeater SAR deceptive jamming, the scene template provides the amplitude modulation map for forwarded echoes. Motivated by image blending, this paper formulates deceptive jamming template construction as an SAR deceptive scene synthesis problem and proposes MGFV-SDTS, a variational optimization framework guided by multi-scale gradients and deep features. The multi-scale Laplacian guidance controls local boundary transitions between the protected region and its background. The feature-guided content and style terms preserve the selected false scene texture and its SAR statistical properties. Experiments on the SAR-Airport-1.0 dataset show that MGFV-SDTS preserves the intended deceptive texture better than representative baselines while maintaining visually acceptable boundary transitions. Ablation results confirm that feature guidance retains false scene content, whereas multi-scale gradients reduce boundary incompatibility. SAR deceptive jamming simulations on the spaceborne dataset and the measured airborne SAR case further support the applicability of the synthesized templates within the tested jamming chain.
Zihan Zhuang, Kai Xie, Jinjian Lin et al.· Electronics· 0 citations
Abstract. Satellite imagery offers a distinct advantage in Earth observation by providing expansive coverage and enabling the monitoring of inaccessible regions without physical on-site intervention, serving as a significantly more cost-effective and scalable alternative to traditional aerial or ground-based surveys. The task of 3D reconstruction from multi-view satellite images has therefore been a pivotal point of research at the intersection of photogrammetry and remote sensing. Recently, novel-view synthesis techniques such as Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS) have accelerated the accuracy and speed of topographic modeling. Among these, Earth Observation Gaussian Splatting (EOGS) has emerged as a state-of-the-art approach by adapting 3DGS to handle the unique geometric and radiometric characteristics of satellite data, including Rational Polynomial Coefficients (RPCs) and varying solar conditions. However, the standard EOGS pipeline relies on stochastic initialization, where Gaussians are distributed uniformly within a volumetric bounding box, leading to high computational overhead and dependency on aggressive pruning that can inadvertently remove critical geometric features, particularly in areas with complex urban structures. To address these limitations, we propose Bundle-Adjusted Initialization for Earth Observation Gaussian Splatting, which leverages sparse point clouds from bundle adjustment as geometric priors for Gaussian initialization. Combined with an adaptive densification strategy, our method achieves faster convergence and improved DSM accuracy on the DFC2019 dataset compared to the EOGS baseline.
Jiyong Kim, Shuang Song, Rongjun Qin· The International Archives o...· 0 citations
Generating synthetic aperture radar (SAR) images at unknown azimuth angles under limited sample conditions remains a challenging task, since target scattering characteristics vary significantly with observation angle and are difficult to model effectively using only image-domain priors. To address this issue, this paper proposes a scattering center prior-guided conditional diffusion framework for unknown-azimuth SAR image generation. First, the Iterative Shrinkage–Thresholding Algorithm (ISTA) constrained by the Point Spread Function (PSF) is employed to extract dominant scattering centers from SAR images at known viewing angles, obtaining sparse and physically interpretable scattering center maps. Subsequently, the target category, azimuth angle, and scattering map are jointly used as conditional vector inputs to train the diffusion model. During the inference stage, scattering maps from known azimuth angles are fused to construct a scattering prior for the unknown azimuth, which is used to guide the generation of the corresponding SAR image. Experimental results under sparse angular sampling conditions demonstrate that, compared with the scattering-guided GAN baseline and the non-learning image-domain interpolation baseline, the proposed method better preserves dominant scattering structures and generates unknown-azimuth SAR images with clearer target contours, more stable strong scattering regions, and fewer local artifacts. In summary, introducing dominant scattering center priors into the conditional diffusion model provides effective physical constraints for SAR image generation and improves unseen-azimuth SAR image generation under the evaluated sparse angular sampling conditions.
