Radar-Based Shape and Reflectivity Reconstruction Using Active Surfaces and the Level Set Method

Area of Science:

• Computer Vision
• 3D Reconstruction
• Radar Imaging

Background:

• Active surface models are used for shape reconstruction.
• Variational methods offer a robust framework for inverse problems.
• Multiview stereo reconstruction typically relies on camera images.

Purpose of the Study:

• To generalize active surface models for multiview shape reconstruction using radar data.
• To adapt variational methods for non-image-based sensor data.
• To explore radar as a single modality for 3D reconstruction.

Main Methods:

• Development of an active surface model driven by sparse radar measurements.
• Adaptation of regularization strategies for radar data within a variational framework.
• Generalization of functional graph-based reconstruction to arbitrary geometries.

Main Results:

• Demonstrated effective 3D shape reconstruction using radar as a single sensor modality.
• Achieved reconstruction quality comparable to traditional image-based methods.
• Identified necessary modifications in regularization for radar data integration.

Conclusions:

• The proposed active surface model effectively reconstructs 3D shapes from radar data.
• This approach shows significant potential for future sensor fusion frameworks.
• Radar-based reconstruction offers a viable alternative to image-based methods.