For oriented normal estimation, previous methods usually conduct a two-stage pipeline, i.e., (1) unoriented normal estimation and (2) normal orientation, while our method achieves this through Neural Gradient Learning (NGL) and Gradient Vector Optimization (GVO). We introduce effective novel designs into our method that enable it to improve the SOTA results.
We propose Neural Gradient Learning (NGL), a deep learning approach to learn gradient vectors with consistent orientation from 3D point clouds for normal estimation. It has excellent gradient approximation properties for the underlying geometry of the data. We utilize a simple neural network to parameterize the objective function to produce gradients at points using a global implicit representation. However, the derived gradients usually drift away from the ground-truth oriented normals due to the lack of local detail descriptions. Therefore, we introduce Gradient Vector Optimization (GVO) to learn an angular distance field based on local plane geometry to refine the coarse gradient vectors. Finally, we formulate our method with a two-phase pipeline of coarse estimation followed by refinement. Moreover, we integrate two weighting functions, i.e., anisotropic kernel and inlier score, into the optimization to improve the robust and detail-preserving performance. Our method efficiently conducts global gradient approximation while achieving better accuracy and generalization ability of local feature description. This leads to a state-of-the-art normal estimator that is robust to noise, outliers and point density variations. Extensive evaluations show that our method outperforms previous works in both unoriented and oriented normal estimation on widely used benchmarks.
Overview of our method. (a-c): The neural gradient learning function f takes a point cloud P as input and derives point-wise gradient ∇f within the network based on neighboring regions of the surface. (d-f): The gradient vector optimization function g selects the optimal vector sample according to angular distance as the normal n.
| Scene 1 | Scene 2 |
|---|---|
|
|
|
@inproceedings{li2023neural,
title={Neural Gradient Learning and Optimization for Oriented Point Normal Estimation},
author={Li, Qing and Feng, Huifang and Shi, Kanle and Fang, Yi and Liu, Yu-Shen and Han, Zhizhong},
booktitle={SIGGRAPH Asia 2023 Conference Papers},
year={2023}
}