Go-ICP: A Globally Optimal Solution to 3D ICP Point-Set Registration
TL;DR: This paper presents the first globally optimal algorithm, named Go-ICP, for Euclidean (rigid) registration of two 3D point-sets under the inline-formula notation, and derives novel upper and lower bounds for the registration error function.
read more
Abstract: The Iterative Closest Point (ICP) algorithm is one of the most widely used methods for point-set registration. However, being based on local iterative optimization, ICP is known to be susceptible to local minima. Its performance critically relies on the quality of the initialization and only local optimality is guaranteed. This paper presents the first globally optimal algorithm, named Go-ICP, for Euclidean (rigid) registration of two 3D point-sets under the $L_2$ error metric defined in ICP. The Go-ICP method is based on a branch-and-bound scheme that searches the entire 3D motion space $SE(3)$ . By exploiting the special structure of $SE(3)$ geometry, we derive novel upper and lower bounds for the registration error function. Local ICP is integrated into the BnB scheme, which speeds up the new method while guaranteeing global optimality. We also discuss extensions, addressing the issue of outlier robustness. The evaluation demonstrates that the proposed method is able to produce reliable registration results regardless of the initialization. Go-ICP can be applied in scenarios where an optimal solution is desirable or where a good initialization is not always available.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Figures
TABLE 1: Running time (in seconds) of Go-ICP with DTs for the registration of the partially overlapping point-sets in Fig. 16. 100 random relative poses were tested for each point-set pair and 1 000 data points were used. ρ is the trimming percentage. 
Fig. 16: Registration with partial overlap. Go-ICP with the trimming strategy successfully registered the 10 point-set pairs with 100 random relative poses for each of them. The point-sets in red and blue are denoted as point-setA and point-setB, respectively. The trimming settings and running times are presented in Table 1. 
Fig. 17: Registration with high optimal error. Left: Gaussian noise was added to the data point-set to increase the RMS error. Right: the global minimum was found at about 25s with a DT; the remainder of the time was devoted solely to increasing the lower bound. 
Fig. 15: Registration with different levels of Gaussian noise. 
Fig. 14: Running time histograms of Go-ICP with DTs for the bunny (left) and dragon (right) point-sets. 
Fig. 13: Running time of the Go-ICP method with DTs on the bunny and dragon point-sets with respect to different factors. The evaluation was conducted on 10 data point-sets with 100 random poses (i.e., 1 000 pairwise registrations).
Citations
MULLS: Versatile LiDAR SLAM via Multi-metric Linear Least Square
Yue Pan,Pengchuan Xiao,Yujie He,Zhenlei Shao,Zesong Li +4 more
- 30 May 2021
TL;DR: Li et al. as mentioned in this paper proposed MULLS, an efficient, low-drift, and versatile 3D LiDAR SLAM system, which uses a multi-metric linear least square iterative closest point algorithm.
199
•Posted Content
PCRNet: Point Cloud Registration Network using PointNet Encoding.
Vinit Sarode,Xueqian Li,Hunter Goforth,Yasuhiro Aoki,Rangaprasad Arun Srivatsan,Simon Lucey,Howie Choset +6 more
TL;DR: A novel framework that uses the PointNet representation to align point clouds and perform registration for applications such as tracking, 3D reconstruction and pose estimation is presented.
194
DeepGMR: Learning Latent Gaussian Mixture Models for Registration
Wentao Yuan,Benjamin Eckart,Kihwan Kim,Varun Jampani,Dieter Fox,Jan Kautz +5 more
- 23 Aug 2020
TL;DR: Deep Gaussian Mixture Registration (DeepGMR) as discussed by the authors is the first learning-based registration method that explicitly leverages a probabilistic registration paradigm by formulating registration as the minimization of KL-divergence between two probability distributions modeled as mixtures of Gaussians.
186
•Posted Content
Feature-metric Registration: A Fast Semi-supervised Approach for Robust Point Cloud Registration without Correspondences
TL;DR: A fast feature-metric point cloud registration framework, which enforces the optimisation of registration by minimising a feature-Metric projection error without correspondences, which obtains higher accuracy and robustness than the state-of-the-art methods.
156
A symmetric objective function for ICP
TL;DR: A new symmetrized objective function is introduced that achieves the simplicity and computational efficiency of point-to-plane optimization, while yielding improved convergence speed and a wider convergence basin.
154
References
Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography
TL;DR: New results are derived on the minimum number of landmarks needed to obtain a solution, and algorithms are presented for computing these minimum-landmark solutions in closed form that provide the basis for an automatic system that can solve the Location Determination Problem under difficult viewing.
A method for registration of 3-D shapes
Paul J. Besl,H.D. McKay +1 more
TL;DR: In this paper, the authors describe a general-purpose representation-independent method for the accurate and computationally efficient registration of 3D shapes including free-form curves and surfaces, based on the iterative closest point (ICP) algorithm, which requires only a procedure to find the closest point on a geometric entity to a given point.
20.6K
•Book
Multiple view geometry in computer vision
Richard Hartley,Andrew Zisserman +1 more
- 01 Jan 2000
TL;DR: In this article, the authors provide comprehensive background material and explain how to apply the methods and implement the algorithms directly in a unified framework, including geometric principles and how to represent objects algebraically so they can be computed and applied.
20.1K
Multiple View Geometry in Computer Vision.
Bernhard P. Wrobel
- 01 Jan 2001
TL;DR: This book is referred to read because it is an inspiring book to give you more chance to get experiences and also thoughts and it will show the best book collections and completed collections.
14.2K
Shape matching and object recognition using shape contexts
TL;DR: This paper presents work on computing shape models that are computationally fast and invariant basic transformations like translation, scaling and rotation, and proposes shape detection using a feature called shape context, which is descriptive of the shape of the object.
7.3K
Related Papers (5)
Radu Bogdan Rusu,Nico Blodow,Michael Beetz +2 more
- 12 May 2009
Szymon Rusinkiewicz,Marc Levoy +1 more
- 01 May 2001