Hybrid Camera Pose Estimation with Online Partitioning

08/05/2019
by   Xinyi Li, et al.
0

This paper presents a hybrid real-time camera pose estimation framework with a novel partitioning scheme and introduces motion averaging to on-line monocular systems. Breaking through the limitations of fixed-size temporal partitioning in most conventional pose estimation mechanisms, the proposed approach significantly improves the accuracy of local bundle adjustment by gathering spatially-strongly-connected cameras into each block. With the dynamic initialization using intermediate computation values, our proposed self-adaptive Levenberg-Marquardt solver achieves a quadratic convergence rate to further enhance the efficiency of the local optimization. Moreover, the dense data association between blocks by virtue of our co-visibility-based partitioning enables us to explore and implement motion averaging to efficiently align the blocks globally, updating camera motion estimations on-the-fly. Experiment results on benchmarks convincingly demonstrate the practicality and robustness of our proposed approach by outperforming conventional bundle adjustment by orders of magnitude.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
08/10/2021

MetaPose: Fast 3D Pose from Multiple Views without 3D Supervision

Recently, huge strides were made in monocular and multi-view pose estima...
research
01/17/2020

Unsupervised Learning of Camera Pose with Compositional Re-estimation

We consider the problem of unsupervised camera pose estimation. Given an...
research
08/24/2019

Camera Pose Correction in SLAM Based on Bias Values of Map Points

Accurate camera pose estimation result is essential for visual SLAM (VSL...
research
02/19/2020

On-line non-overlapping camera calibration net

We propose an easy-to-use non-overlapping camera calibration method. Fir...
research
11/02/2020

Pushing the Envelope of Rotation Averaging for Visual SLAM

As an essential part of structure from motion (SfM) and Simultaneous Loc...
research
06/27/2023

PoseDiffusion: Solving Pose Estimation via Diffusion-aided Bundle Adjustment

Camera pose estimation is a long-standing computer vision problem that t...
research
05/19/2019

Low-latency Visual SLAM with Appearance-Enhanced Local Map Building

A local map module is often implemented in modern VO/VSLAM systems to im...

Please sign up or login with your details

Forgot password? Click here to reset