Chapter 25: Implicit Geometry from RF

Learning Objectives

• Define signed distance functions and verify the Eikonal equation for geometric primitives
• Implement sphere tracing for efficient ray-surface intersection with SDFs
• Describe the GeRaF pipeline from matched-filter power images to SDF-based 3D reconstruction
• Compare occupancy networks with SDF representations for RF scene reconstruction
• Apply Eikonal regularization to enforce valid SDF geometry during neural network training
• Formulate joint estimation of geometry and material properties from multi-view RF data