Multi-scale Feature Line Extraction from Raw Point Clouds Based on Local Surface Variation and Anisotropic Contraction

Introduction

This repository is for our IEEE Transactions on Automation Science and Engineering (TASE) 2021 paper Multi-scale Feature Line Extraction from Raw Point Clouds Based on Local Surface Variation and Anisotropic Contraction.

Usage

Please refer to the Readme.pdf

Related work and dataset

See another learning-based version, which could further enhance the edge extraction accuracy. We have also developed a new dataset containing diverse aircraft panels for evaluating edge detection. This dataset is available from here.

Note that users can run visit_data.m to recover each patch to the original shape space.

Citation

If you use this dataset, please consider citing our work.

@article{chen2024thin,
  title={Thin-walled Aircraft Panel Edge Extraction from 3D Measurement Surfaces via Feature-aware Displacement Learning},
  author={Chen, Mengqi and Zhou, Laishui and Chen, Honghua and Wang, Jun},
  journal={IEEE Transactions on Instrumentation and Measurement},
  year={2024},
  publisher={IEEE}
}

@article{chen2021multiscale,
  title={Multiscale feature line extraction from raw point clouds based on local surface variation and anisotropic contraction},
  author={Chen, Honghua and Huang, Yaoran and Xie, Qian and Liu, Yuanpeng and Zhang, Yuan and Wei, Mingqiang and Wang, Jun},
  journal={IEEE Transactions on Automation Science and Engineering},
  volume={19},
  number={2},
  pages={1003--1016},
  year={2021},
  publisher={IEEE}
}