Loop Closing for Visual Pose Tracking during Close-Range 3-D Modeling

Abstract

This work deals with the passive tracking of the pose of a close-range 3-D modeling device using its own high-rate images in realtime, concurrently with customary 3-D modeling of the scene by laser triangulation. Our former works in Refs. [1,2] successfully implemented visual pose tracking. Accuracy being a central requirement to 3-D modeling, however, here we note that accuracy can be further increased using a graph-based nonlinear optimization of the tracked pose by minimization of reprojection errors. Loop closures e.g. when having scanned all around the objects provide the opportunity to increase pose tracking and 3-D modeling accuracy. The sparse optimization is in the form of a hybrid, keyframe-based bundle adjustment algorithm on stereo keyframes, yielding rapid optimization of the whole trajectory and object mesh model within a second. The optimization is supported by the use of appearance-based SURF descriptors together with a bank of parallel three-point-perspective pose solvers.