點云(剛性)配準(zhǔn)，求解兩個(具有overlap的)點云P, Q之間的變換(旋轉(zhuǎn)矩陣和平移向量)，使得點云P, Q對齊(可以通俗的理解為使P, Q的坐標(biāo)處于同一坐標(biāo)系下)。點云配準(zhǔn)在無人駕駛、三維重建等領(lǐng)域具有廣泛的應(yīng)用。

本文整理了點云配準(zhǔn)相關(guān)的論文，既包括基于深度學(xué)習(xí)的點云配準(zhǔn)算法，也包括部分傳統(tǒng)配準(zhǔn)算法(ICP, GoICP, FGR等)。先上一張點云配準(zhǔn)的發(fā)展脈絡(luò)圖Figure 1。但是需要注意兩點: (1) Figure 1中列出了方法并不全面，僅選取了部分代表性的點云配準(zhǔn)方法, 更多論文請參考下面的列表; (2) Figure 1中記錄的同一年內(nèi)的配準(zhǔn)方法的提出順序可能有錯誤。

• Point Cloud Registration using Representative Overlapping Points [arXiv 2021; PyTorch]
• Shape registration in the time of transformers [arXiv 2021]
• 3D Shape Registration Using Spectral Graph Embedding and Probabilistic Matching [arXiv 2021]
• FINet: Dual Branches Feature Interaction for Partial-to-Partial Point Cloud Registration [arXiv 2021]
• Generalisable and distinctive 3D local
  deep descriptors for point cloud registration [arXiv 2021]
• Deep Weighted Consensus (DWC) Dense correspondence confidence maps for 3D shape registration [arXiv 2021]
• 3D Point Cloud Registration with Multi-Scale Architecture and Self-supervised Fine-tuning [arXiv 2021; Github]
• R-PointHop: A Green, Accurate and Unsupervised Point Cloud Registration Method [arXiv 2021; Github]
• OMNet: Learning Overlapping Mask for Partial-to-Partial Point Cloud Registration [arXiv 2021]
• UnsupervisedR&R: Unsupervised Point Cloud Registration via Differentiable Rendering [arXiv 2021; PyTorch]
• PREDATOR: Registration of 3D Point Clouds with Low Overlap [CVPR 2021; PyTorch]
• SpinNet: Learning a General Surface Descriptor for 3D Point Cloud Registration [CVPR 2021; Github]
• PointDSC: Robust Point Cloud Registration using Deep Spatial Consistency [CVPR 2021; PyTorch]
• Robust Point Cloud Registration Framework Based on Deep Graph Matching [CVPR 2021; Github]
• Fast and Robust Iterative Closest Point [TPAMI 2021; Github]
• RPSRNet: End-to-End Trainable Rigid Point Set Registration Network using Barnes-Hut 2D-Tree Representation [CVPR 2021]
• ReAgent: Point Cloud Registration using Imitation and Reinforcement Learning [CVPR 2021; PyTorch]
• Deep Global Registration [CVPR 2020; PyTorch]
• 3DRegNet: A Deep Neural Network for 3D Point Registration [CVPR 2020; Tensorflow]
• D3Feat: Joint Learning of Dense Detection and Description of 3D Local Features [CVPR 2020; Tensorflow, PyTorch]
• Feature-metric Registration: A Fast Semi-supervised Approach for Robust Point Cloud Registration without Correspondences [CVPR 2020; PyTorch]
• RPM-Net: Robust Point Matching using Learned Features [CVPR 2020; PyTorch]
• Learning multiview 3D point cloud registration [CVPR 2020; PyTorch]
• DeepGMR: Learning Latent Gaussian Mixture Models for Registration [ECCV 2020; PyTorch]
• Iterative Distance-Aware Similarity Matrix Convolution with Mutual-Supervised Point Elimination for Efficient Point Cloud Registration [ECCV 2020; PyTorch]
• Fast and Automatic Registration of Terrestrial Point Clouds Using 2D Line Features [Remote Sensing 2020]
• PCRNet: Point Cloud Registration Network using PointNet Encoding [arXiv 2020; PyTorch, Tensorflow]]
• TEASER: Fast and Certifiable Point Cloud Registration [arXiv 2020; Github]
• PRNet: Self-Supervised Learning for Partial-to-Partial Registration [NeurIPS 2019; PyTorch]
• Deep Closest Point: Learning Representations for Point Cloud Registration [ICCV 2019; PyTorch]
• DeepICP: An End-to-End Deep Neural Network for 3D Point Cloud Registration [ICCV 2019]
• USIP: Unsupervised Stable Interest Point Detection from 3D Point Clouds [ICCV 2019; PyTorch]
• 3D Local Features for Direct Pairwise Registration [CVPR 2019]
• DeepMapping: Unsupervised Map Estimation From Multiple Point Clouds [CVPR 2019; Github]
• PointNetLK: Point Cloud Registration using PointNet [CVPR 2019; PyTorch]
• PPFNet: Global Context Aware Local Features for Robust 3D Point Matching [CVPR 2018]
• 3DFeat-Net: Weakly Supervised Local 3D Features for Point Cloud Registration [ECCV 2018; Tensorflow]
• PPFNet: Global Context Aware Local Features for Robust 3D Point Matching [ECCV 2018]
• Iterative Global Similarity Points : A robust coarse-to-fine integration solution for pairwise 3D point cloud registration [3DV 2018]
• Guaranteed Outlier Removal for Point Cloud Registration with Correspondences [TPAMI 2018]
• 3DMatch: Learning Local Geometric Descriptors from RGB-D Reconstructions [CVPR 2017; project]
• 3D Point Cloud Registration for Localization using a Deep Neural Network Auto-Encoder [CVPR 2017; github]
• Fast Global Registration [ECCV 2016; Github]
• Go-ICP: A Globally Optimal Solution to 3D ICP Point-Set Registration [TPAMI 2015; Github]
• Fast point feature histograms (FPFH) for 3D registration [ICRA 2009]
• A method for registration of 3-D shapes [TPAMI 1992]
• Least-squares fitting of two 3-D point sets [TPAMI 1987]

更多點云相關(guān)(分類、分割、檢測等)的文章列表詳見https://github.com/zhulf0804/3D-PointCloud。

最后，安利一下自己的基于深度學(xué)習(xí)的點云配準(zhǔn)工作Point Cloud Registration using Representative Overlapping Points, 代碼也已經(jīng)開源, https://github.com/zhulf0804/ROPNet。歡迎大家一塊交流學(xué)習(xí)~

總結(jié)

以上是生活随笔為你收集整理的点云配准论文的全部內(nèi)容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網(wǎng)站內(nèi)容還不錯，歡迎將生活随笔推薦給好友。