Capturing Dynamic Textured Surfaces of Moving Targets
This enables dynamic 3D reconstruction for applications like animation or virtual reality without expensive equipment or templates, though it appears incremental as it builds on existing registration and reconstruction approaches.
The paper tackles the problem of reconstructing complete 3D models of moving subjects like humans and animals using only a few handheld sensors, achieving reliable registration with as little as 15% overlap between scans and outperforming alternative methods.
We present an end-to-end system for reconstructing complete watertight and textured models of moving subjects such as clothed humans and animals, using only three or four handheld sensors. The heart of our framework is a new pairwise registration algorithm that minimizes, using a particle swarm strategy, an alignment error metric based on mutual visibility and occlusion. We show that this algorithm reliably registers partial scans with as little as 15% overlap without requiring any initial correspondences, and outperforms alternative global registration algorithms. This registration algorithm allows us to reconstruct moving subjects from free-viewpoint video produced by consumer-grade sensors, without extensive sensor calibration, constrained capture volume, expensive arrays of cameras, or templates of the subject geometry.