MicroG

《Micro-Structures Graph-Based Point Cloud Registration for Balancing Efficiency and Accuracy》[arxiv]

News

• [2024-11-17]: Early Access
• [2024-10-26]: Our paper has been accpeted by TGRS (IEEE Transactions on Geoscience and Remote Sensing).

Table of Contents

1. Introduction
2. Dependencies
3. Usage
4. Data-preparation
5. Acknowledgments
6. Citation

Introduction

This paper introduces a micro-structures graph-based coarse-to-fine global point cloud registration method. This method employs a hierarchical outlier removal strategy based on graph nodes and edges, combined with the GNC-Welsch estimator, to ensure robustness during coarse registration. At finer scales, PA-AA optimization is utilized to further exploit the geometric features of corresponding micro-structures, enhancing accuracy with minimal additional computational cost.

We have conducted experiments on ETH and 3DMatch dataset.

Dependencies

We've tested it on CLion 2024 running on Ubuntu 18.04.
A compiler that supports OpenMP.

• CMake >= 3.10
• PCL >=1.12
• Eigen3 >=3.3.0

Usage

1. Clone the repository:

   git clone [email protected]:Rolin-zrl/MicroG.git

2. Compile

   mkdir build
   cd build
   cmake ..
   make
   ./ MicroG "../data/s1.ply" "../data/s1.ply" "./data/s1-s2.tfm" 0.1 800 2
   

Input six parameters introduced in the paper

1. source_path
2. target_path
3. gt_path (optional)
4. resolution: downsampling resolution
5. K: default 800
6. L: default 2.0

Data-preparation

You can test on the online available point cloud data and registration dataset, such as
WHU TLS Registration Dataset,
ETH PRS TLS Registration Dataset,
ETH ASL Robotics Registration Dataset,
3D Match,
Robotic 3D Scan Repository, etc.

Acknowledgments

People who inspired this idea, gave suggestions and reported errors, especially Pengcheng Wei contributing to the removal of outliers.

Citation

@ARTICLE{10755047,
author={Zhang, Rongling and Yan, Li and Wei, Pengcheng and Xie, Hong and Wang, Pinzhuo and Wang, Binbing},
journal={IEEE Transactions on Geoscience and Remote Sensing},
title={Micro-Structures Graph-Based Point Cloud Registration for Balancing Efficiency and Accuracy},
year={2024},
volume={},
number={},
pages={1-1},
keywords={Point cloud registration;correspondence graph;robust estimator;planar adjustment;Anderson acceleration},
doi={10.1109/TGRS.2024.3488502}}