Point Cloud Registration是指,为得到被测物体的完整数据模型,需确定一个合适的坐标变换,将从各个视角得到的点集合并到一个统一的坐标系下,形成一个完整的数据点云,以方便地进行可视化等操作。
We propose the first fast and certifiable algorithm for the registration of two sets of 3D points in the presence of large amounts of outlier correspondences
Cited by19BibtexViews11DOI
0
0
IEEE Robotics Autom. Lett., no. 2 (2020): 1127-1134
The approach allows extending the applicability of Black-Rangarajan duality and Graduated Non-convexity to several spatial perception problems, ranging from mesh registration and shape alignment to pose graph optimization
Cited by8BibtexViews18DOI
0
0
CVPR, pp.7191-7201, (2020)
We propose 3DRegNet, a deep neural network that can solve the scan registration problem by jointly solving the outlier rejection given 3D point correspondences and computing the pose for alignment of the scans
Cited by6BibtexViews39DOI
0
0
NeurIPS, (2020)
We have proposed a general framework for designing certifiable algorithms for a broad class of robust geometric perception problems
Cited by2BibtexViews15DOI
0
0
CVPR, pp.1756-1766, (2020)
Research on local descriptors for pairwise registration of 3D point clouds is centered on deep learning approaches that succeed in capturing and encoding evidence hidden to hand-engineered descriptors
Cited by2BibtexViews69DOI
0
0
CVPR, pp.12051-12060, (2020)
We have introduced a novel framework to transform point clouds into a hierarchical Gaussian mixture model representation
Cited by1BibtexViews40DOI
0
0
CVPR, pp.11283-11292, (2020)
We have proposed Attentive Context Normalization, and used it to build Attentive Context Networks to solve problems on permutation-equivariant data
Cited by1BibtexViews31DOI
0
0
CVPR, pp.6358-6366, (2020)
The registration is considered accurate if the Relative Translation Error is below 2m and Relative Rotation Error is below 5◦ following
Cited by1BibtexViews28DOI
0
0
CVPR, pp.11363-11371, (2020)
We propose a feature-metric framework to solve the point cloud registration, and the framework can be trained using a semi-supervised or unsupervised manner
Cited by0BibtexViews35DOI
0
0
CVPR, pp.1916-1925, (2020)
We have presented a novel end-to-end framework for learning local multi-view descriptors of 3D point clouds
Cited by0BibtexViews44DOI
0
0
Shengyu Huang,Zan Gojcic, Mikhail Usvyatsov, Andreas Wieser,Konrad Schindler
The core of the model is an overlap attention module that enables early information exchange between the point clouds’ latent encodings, in order to infer which of their points are likely to lie in their overlap region
Cited by0BibtexViews3DOI
0
0
Extracting robust and general 3D local features is key to downstream tasks such as point cloud registration and reconstruction. Existing learning-based local descriptors are either sensitive to rotation transformations, or rely on classical handcrafted features which are neithe...
Cited by0BibtexViews4DOI
0
0
We presented CLIPPER, a graph-theoretic framework for robust data association using the notion of geometric consistency
Cited by0BibtexViews1DOI
0
0
Jingnan Shi, Heng Yang,Luca Carlone
This paper develops an approach to prune outliers
Cited by0BibtexViews3DOI
0
0
Felix Järemo Lawin, Per-Erik Forssén
We have extended the paradigm of probabilistic point set registration to exploit the discriminative powers of learned features and weights
Cited by0BibtexViews0DOI
0
0
Vinit Sarode, Animesh Dhagat,Rangaprasad Arun Srivatsan, Nicolas Zevallos,Simon Lucey,Howie Choset
That MaskNet – augments the ability of existing classical and deep learning-based registration methods to better deal with partial point clouds and outliers, can be used to reject noise, and generalizes to object categories that it was not trained on
Cited by0BibtexViews4DOI
0
0
While most competing methods get trapped in the local minima caused by non-uniform point sampling, our Fast Gravitational Approach recovers transformations which are closer to the optimal ones, thanks to the boundary conditions defined on masses via the Smooth-Particle Mass funct...
Cited by0BibtexViews9DOI
0
0
Fabio Poiesi, Davide Boscaini
We presented a novel approach to learn local, compact and rotation invariant descriptors end-to-end through a PointNetbased deep neural network using canonicalised patches
Cited by0BibtexViews7DOI
0
0
european conference on computer vision, pp.733-750, (2020)
We have proposed Deep Gaussian Mixture Registration, a first attempt towards learning-based probabilistic registration
Cited by0BibtexViews19DOI
0
0
Jiaqi Yang, Jiahao Chen, Zhiqiang Huang, Siwen Quan,Yanning Zhang,Zhiguo Cao
We find that the pipeline of our proposed compatibility feature-based 3D correspondence grouping method can be generalized to matching problems for many other data representations, such as 2D images and non-rigid point clouds/meshes, which remains an interesting future research d...
Cited by0BibtexViews37DOI
0
0