Sparse 3D Point-cloud Map Upsampling and Noise Removal as a vSLAM Post-processing Step: Experimental Evaluation

TL;DR

This paper evaluates post-processing techniques for sparse 3D point-cloud maps from vSLAM, focusing on noise removal and upsampling, and demonstrates their application in converting UAV maps for path planning.

Contribution

It systematically compares known algorithms for noise removal and upsampling in vSLAM maps and identifies the most effective combination for real-world indoor and outdoor environments.

Findings

Optimal algorithm combination for noise removal and upsampling identified.

Converted UAV point-cloud maps into 3D voxel grids suitable for path planning.

Demonstrated effectiveness in real indoor flight scenarios.

Abstract

The monocular vision-based simultaneous localization and mapping (vSLAM) is one of the most challenging problem in mobile robotics and computer vision. In this work we study the post-processing techniques applied to sparse 3D point-cloud maps, obtained by feature-based vSLAM algorithms. Map post-processing is split into 2 major steps: 1) noise and outlier removal and 2) upsampling. We evaluate different combinations of known algorithms for outlier removing and upsampling on datasets of real indoor and outdoor environments and identify the most promising combination. We further use it to convert a point-cloud map, obtained by the real UAV performing indoor flight to 3D voxel grid (octo-map) potentially suitable for path planning.