Robust Rotation Synchronization via Low-rank and Sparse Matrix Decomposition

Abstract

This paper deals with the rotation synchronization problem, which arises in global registration of 3D point-sets and in structure from motion. The problem is formulated in an unprecedented way as a "low-rank and sparse" matrix decomposition that handles both outliers and missing data. A minimization strategy, dubbed R-GoDec, is also proposed and evaluated experimentally against state-of-the-art algorithms on simulated and real data. The results show that R-GoDec is the fastest among the robust algorithms.

The material contained in this paper is part of a manuscript submitted to CVIU

References in corpus (5)

• Spectral Motion Synchronization in SE(3)
• Semidefinite descriptions of the convex hull of rotation matrices
• A Solution for Multi-Alignment by Transformation Synchronisation
• On Computing the Translations Norm in the Epipolar Graph
• ShapeFit and ShapeKick for Robust, Scalable Structure from Motion

Cited by in corpus (10)

• Modeling Perceptual Aliasing in SLAM via Discrete-Continuous Graphical Models
• Registration of multi-view point sets under the perspective of expectation-maximization
• RAGO: Recurrent Graph Optimizer For Multiple Rotation Averaging
• Message Passing Least Squares Framework and its Application to Rotation Synchronization
• Learning to Communicate and Correct Pose Errors
• Learning Transformation Synchronization
• 3DMNDT:3D multi-view registration method based on the normal distributions transform
• Solving Viewing Graph Optimization for Simultaneous Position and Rotation Registration
• Effective multi-view registration of point sets based on student's t mixture model
• Path-Invariant Map Networks