Capturing Dynamic Textured Surfaces of Moving Targets

TL;DR

This paper introduces a system that reconstructs complete, textured 3D models of moving subjects using minimal sensors and a novel registration algorithm that handles partial scans and occlusions effectively.

Contribution

The paper presents a new pairwise registration algorithm based on mutual visibility and occlusion, enabling accurate reconstruction of moving subjects from minimal sensor data.

Findings

Reliable registration with as little as 15% overlap

Outperforms existing global registration algorithms

Reconstructs moving subjects without extensive calibration or templates

Abstract

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.