GeoDiff-SAR: A Geometric Prior Guided Diffusion Model for SAR Image Generation

2026-01-07

AI Summary

Existing SAR image generation methods often lack explicit modeling of geometric priors, making it difficult to precisely control critical imaging parameters such as azimuth angle, which limits generation fidelity. To address this, this work proposes GeoDiff-SAR, the first approach to explicitly integrate physical geometric priors into a diffusion model. By leveraging SAR point cloud geometry, a FiLM-based multimodal feature fusion gating mechanism, and LoRA fine-tuning of the Stable Diffusion 3.5 architecture, GeoDiff-SAR enables high-fidelity and controllable SAR image synthesis. Experimental results on real-world datasets demonstrate that the proposed method significantly improves image quality and enhances target recognition accuracy across varying azimuth angles, thereby validating the effectiveness and practicality of physics-guided generative modeling for SAR imagery.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Synthetic Aperture Radar (SAR) imaging results are highly sensitive to observation geometries and the geometric parameters of targets. However, existing generative methods primarily operate within the image domain, neglecting explicit geometric information. This limitation often leads to unsatisfactory generation quality and the inability to precisely control critical parameters such as azimuth angles. To address these challenges, we propose GeoDiff-SAR, a geometric prior guided diffusion model for high-fidelity SAR image generation. Specifically, GeoDiff-SAR first efficiently simulates the geometric structures and scattering relationships inherent in real SAR imaging by calculating SAR point clouds at specific azimuths, which serves as a robust physical guidance. Secondly, to effectively fuse multi-modal information, we employ a feature fusion gating network based on Feature-wise Linear Modulation (FiLM) to dynamically regulate the weight distribution of 3D physical information, image control parameters, and textual description parameters. Thirdly, we utilize the Low-Rank Adaptation (LoRA) architecture to perform lightweight fine-tuning on the advanced Stable Diffusion 3.5 (SD3.5) model, enabling it to rapidly adapt to the distribution characteristics of the SAR domain. To validate the effectiveness of GeoDiff-SAR, extensive comparative experiments were conducted on real-world SAR datasets. The results demonstrate that data generated by GeoDiff-SAR exhibits high fidelity and effectively enhances the accuracy of downstream classification tasks. In particular, it significantly improves recognition performance across different azimuth angles, thereby underscoring the superiority of physics-guided generation.

Problem

• SAR image generation
• geometric prior
• azimuth angle control
• image fidelity
• physical guidance

Innovation

• geometric prior
• diffusion model
• SAR image generation
• FiLM
• LoRA