FootCapture: Towards an AR-based System for 3D Foot Object Acquisition through Photogrammetry
Medical Imaging with Deep Learning, 2024
Key contributions
- An AR dome-based capture interface that guides untrained users to take photogrammetry-optimal foot images on a standard phone.
- More robust and accurate 3D reconstructions than Apple's GuidedCapture in a comparative user study, at comparable usability.
- A flexible, low-cost workflow for clinical uses such as chronic-wound monitoring and orthopedics.
How it works
We present FootCapture, an AR-based mobile application designed to simplify the acquisition of high-quality 3D foot models for clinical applications such as chronic wound monitoring and orthopedics. We developed an intuitive dome-based interface that guides untrained users to capture optimal images for photogrammetry-based reconstruction. In a comparative user study (n=7), we evaluated FootCapture against Apple’s GuidedCapture. While usability scores were comparable, we observed that FootCapture consistently produced more robust and accurate 3D models. Our method demonstrates superior resilience to user errors and enables a flexible, low-cost workflow, making it a valuable tool for clinical practice.
Citation
Khan-Blouki, Valentin; Seiz, Franziska; Walter, Nicolas; Jaus, Alexander; Marinov, Zdravko; Luijten, Gijs; Egger, Jan; Seibold, Constantin Marc; Solte, Dirk; Kleesiek, Jens; Stiefelhagen, Rainer. (2024). "FootCapture: Towards an AR-based System for 3D Foot Object Acquisition through Photogrammetry." Medical Imaging with Deep Learning