Parameter-adaptive multi-frame joint pose optimization method

Camera pose optimization is the basis of geometric vision works, such as 3D reconstruction, structure from motion, and visual odometry. We designed a multi-frame pose optimization method based on the inverse compositional algorithm. The neural networks are added into the optimization model to improv...

Full description

Saved in:
Bibliographic Details
Published in:The Visual computer Vol. 39; no. 7; pp. 2529 - 2541
Main Authors: Li, Shaopeng, Xian, Yong, Wu, Wei, Zhang, Tao, Li, Bangjie
Format: Journal Article
Language:English
Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.07.2023
Springer Nature B.V
Subjects:
Accuracy
Algorithms
Artificial Intelligence
Cameras
Computer Graphics
Computer Science
Frame design
Image Processing and Computer Vision
Image reconstruction
Methods
Multilayers
Neural networks
Optimization
Optimization models
Original Article
Pose estimation
Supervision
Pose optimization
Neural network
Geometric vision
Image process
Machine learning
ISSN:0178-2789, 1432-2315
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Camera pose optimization is the basis of geometric vision works, such as 3D reconstruction, structure from motion, and visual odometry. We designed a multi-frame pose optimization method based on the inverse compositional algorithm. The neural networks are added into the optimization model to improve the problems of hyperparameter selection and loss function design. The multi-frame joint is used to fully utilize the constraints between the sequence images. A multi-layer stepwise method is used, which incorporates scale factors on the loss of each layer to enhance the convergence of the network. The simulation verifies that the proposed method achieves higher precision of pose estimation compared with the state-of-the-art.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0178-2789
1432-2315
DOI:10.1007/s00371-022-02476-4