Physics guided method generates realistic underground radar images
PCFlow: Physics-Conditioned Flow Matching for GPR B-Scan Image Synthesis
Machine Learning
Summary
Ground-penetrating radar (GPR) images help see underground objects, but making fake radar images that look real and follow physics is hard. The authors created PCFlow, a system that uses physics knowledge about materials, shapes, and signals to guide image creation, making the radar images match real-world physics better. They tested PCFlow on radar images of buried pipes and found it makes more accurate and realistic images than previous methods. This helps in making better data for training and testing radar systems.
What this means in practice
- •For radar system engineers: Generate accurate synthetic radar images to test and improve underground object detection algorithms with physical realism.
- •For training data teams: Create more realistic and controllable augmented datasets for machine learning models that analyze ground-penetrating radar scans.
Authors
Zhijie Shen, Chenchen Fu, Xuanhao Chang, Hongtao Bai, Lili He
Abstract
Ground-penetrating radar (GPR) B-scan image synthesis is important for data augmentation, algorithm validation, and simulation acceleration, yet generating radargrams with both visual realism and physical consistency remains challenging. Existing learning-based generative models often emphasize visual appearance but provide limited control over response geometry. In this paper, we propose PCFlow, a physics-conditioned flow matching framework for fast GPR B-scan image synthesis. The core of PCFlow is a Maxwell-informed dense physical condition field constructed from the parameterized physical model used for electromagnetic simulation, including material properties, target geometry, propagation cues, and response-domain priors. This condition field provides an interpretable interface between physical scene parameters and radar response geometry, and guides conditional flow matching in the VAE latent space toward physically feasible generation paths. We evaluate PCFlow on a gprMax-based buried-pipeline dataset with both in-distribution and out-of-distribution test cases. Experimental results show that PCFlow generates images with more accurate response geometry and high visual fidelity, demonstrating its effectiveness for controllable and physically faithful radar image synthesis.