HERE-release

Implementation of the paper:

[Efficient and Robust Point Cloud Registration via Heuristics-guided Parameter Search] (T-PAMI 2024).

To apply this algorithm to your data, you may need to adjust the noise_bound (Line-42 in ./demo.cpp) and BRANCH_ACCURACY (Line-22 in ./include/utils.h). For example, smaller BRANCH_ACCURACY leads to higher accuracy but requires more time cost.

Citation

@article{Huang_2024_TPAMI,
    author       = {Huang, Tianyu and Li, Haoang and Peng, Liangzu and Liu, Yinlong and Liu, Yun-Hui},
    title        = {Efficient and Robust Point Cloud Registration via Heuristics-guided Parameter Search},
    journal      = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
    year         = {2024}
    publisher={IEEE}
}

Our another work regarding 3D registration

Scalable 3D Registration via Truncated Entry-wise Absolute Residuals (CVPR 2024)

Acknowledgements

The Branch-and-Bound algorithm part of our code is built on GORE. We thank the authors for their excellent work!