Research map / Point clouds / Registration and matching
Joint Alignment of Multiple Point Sets with Batch and Incremental Expectation-Maximization
Georgios Evangelidis, Radu Horaud
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use joint registration with a common GMM (JRMPC-B) instead of one-vs-all pairwise methods | ICP, CPD, GMMReg, SimReg · simulated Bunny/Lucy/Armadillo (SNR=10dB, 30% outliers) | indirect rotation error std: 0.047 -> 0.008 (Bunny vs ICP) | Improved |
| use JRMPC-B instead of MATrICP | MATrICP · Dragon without noise | rotation error (degrees): 0.62 -> 0.47 | Improved |
| use JRMPC-B instead of MATrICP | MATrICP · Bunny without noise | rotation error (degrees): 0.27 -> 0.37 | Hurt |
| use JRMPC-B instead of MATrICP with SNR 25dB noise | MATrICP · Bunny with noise | rotation error (degrees): 2.37 -> 1.07 | Improved |
| use JRMPC-B instead of MAICP with noise | MAICP · Happy Buddha with noise | rotation error (degrees): 4.13 -> 1.18 | Improved |
| use batch JRMPC-B instead of incremental JRMPC-I | JRMPC-I · Bunny without noise | rotation error (degrees): 0.69 -> 0.37 | Improved |
| use incremental JRMPC-I instead of motion averaging under noise | MAICP, MATrICP · Dragon with noise | rotation error (degrees): 1.55 -> 0.89 | Improved |
| use fixed matches (MV-ICP) | JRMPC · multi-view registration | rotation error: 3.75 on Dragon without noise | Hurt |
| fix the variance for initial iterations (ablation) | JRMPC-B · multi-view registration | convergence level | Improved |
| keep priors constant instead of updating (ablation) | JRMPC · simulated data | registration quality | No clear effect |
| use JRMPC-I with hierarchical scheme instead of SLAM method on fr2/desk | Elastic fusion, RGBD SLAM · TUM fr2/desk | translation RMSE (m): 0.071 -> 0.034 | Improved |
| use JRMPC-I instead of ORB-SLAM2 | ORB-SLAM2 · TUM fr1/desk | translation RMSE (m): 0.016 -> 0.047 | Hurt |
Closest papers by meaning
- Probabilistic Model for Robust Affine and Non-Rigid Point Set Matching (TPAMI 2016 · similarity 0.89)
- Group-Wise Point-Set Registration Based on Renyi's Second Order Entropy (CVPR 2017 · similarity 0.88)
- GOGMA: Globally-Optimal Gaussian Mixture Alignment (CVPR 2016 · similarity 0.88)
- Efficient Global Point Cloud Alignment Using Bayesian Nonparametric Mixtures (CVPR 2017 · similarity 0.87)
- Correspondence-Free Non-Rigid Point Set Registration Using Unsupervised Clustering Analysis (CVPR 2024 · similarity 0.86)
- A robust inlier identification algorithm for point cloud registration via $\mathbf{\ell_0}$-minimization (NeurIPS 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
