A Complex-valued SAR Foundation Model Based on Physically Inspired Representation Learning

2025-04-16

AI Summary

To address insufficient information utilization and poor interpretability in SAR image foundation models, this paper proposes the first physics-driven foundation model for complex-valued SAR imagery. Methodologically, it innovatively embeds polarization decomposition—a physically grounded scattering process—into a self-supervised pretraining framework: pixel-wise scattering intensity is modeled via a learnable weighted combination of scattering bases and physically interpretable Yamaguchi decomposition coefficients; the architecture incorporates complex-valued neural networks, a scattering-query decoder, and dual losses—polarimetric decomposition loss (enforcing alignment with Yamaguchi coefficients) and complex power reconstruction loss. Our key contributions include the first integration of scattering physics into model design, learnable scattering bases, physically meaningful and interpretable coefficients, and latent representations with explicit electromagnetic semantics—significantly enhancing model transparency and few-shot generalization. The model achieves state-of-the-art performance across six downstream tasks, demonstrating robust and generalizable feature representations under data scarcity.

Technology Category

• Computer Vision: Low Level & Physics-based Vision
• Machine Learning: Deep Neural Architectures and Foundation Models
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

Vision foundation models in remote sensing have been extensively studied due to their superior generalization on various downstream tasks. Synthetic Aperture Radar (SAR) offers all-day, all-weather imaging capabilities, providing significant advantages for Earth observation. However, establishing a foundation model for SAR image interpretation inevitably encounters the challenges of insufficient information utilization and poor interpretability. In this paper, we propose a remote sensing foundation model based on complex-valued SAR data, which simulates the polarimetric decomposition process for pre-training, i.e., characterizing pixel scattering intensity as a weighted combination of scattering bases and scattering coefficients, thereby endowing the foundation model with physical interpretability. Specifically, we construct a series of scattering queries, each representing an independent and meaningful scattering basis, which interact with SAR features in the scattering query decoder and output the corresponding scattering coefficient. To guide the pre-training process, polarimetric decomposition loss and power self-supervision loss are constructed. The former aligns the predicted coefficients with Yamaguchi coefficients, while the latter reconstructs power from the predicted coefficients and compares it to the input image's power. The performance of our foundation model is validated on six typical downstream tasks, achieving state-of-the-art results. Notably, the foundation model can extract stable feature representations and exhibits strong generalization, even in data-scarce conditions.

Problem

• Develops complex-valued SAR foundation model for better Earth observation
• Addresses insufficient information use and poor interpretability in SAR models
• Enhances model performance with physics-inspired pre-training and decomposition

Innovation

• Complex-valued SAR model with physical interpretability
• Scattering queries simulate polarimetric decomposition
• Polarimetric and power loss guide pre-training