An unsupervised approach for robust point cloud registration with deep feature

Shenxgi Wei; Ming Chen; Shenglian Lu; Weijie Bi

doi:10.1117/12.3029999

8 May 2024 An unsupervised approach for robust point cloud registration with deep feature

Shenxgi Wei, Ming Chen, Shenglian Lu, Weijie Bi

Proceedings Volume 13162, Fourth Symposium on Pattern Recognition and Applications (SPRA 2023); 1316207 (2024) https://doi.org/10.1117/12.3029999
Event: Fourth Symposium on Pattern Recognition and Applications (SPRA2023), 2023, Napoli, Italy

Abstract

We propose a point cloud registration method based on deep learning, which achieves the purpose of point cloud registration by reducing the projection error of point cloud in feature space. When the two point clouds are basically aligned, the projection mapping of their deep features is also highly similar. Our framework has two pipeline branches. The main branch aims to register the point cloud. Its novelty lies in discarding the traditional calculation of point pair relations, mapping the point cloud into a deep feature map, and achieving registration by reducing the projection error of the two maps. The sub pipeline is a "teacher" who mainly trains the encoder, so that the model can be trained in an unsupervised way without expensive data marking. And compared with some recent unsupervised methods, our method does not only rely on global descriptors, but also attaches importance to the extraction of robust local feature descriptors. Experiments show the state-of-the-art performance of our method which is joint extracting high-level global and local representations in an unsupervised manner, requiring no labeled data or arduously searching correspondences.

(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Shenxgi Wei, Ming Chen, Shenglian Lu, and Weijie Bi "An unsupervised approach for robust point cloud registration with deep feature", Proc. SPIE 13162, Fourth Symposium on Pattern Recognition and Applications (SPRA 2023), 1316207 (8 May 2024); https://doi.org/10.1117/12.3029999

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
10 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Point clouds

Education and training

Matrices

Feature extraction

Image registration

3D modeling

Clouds

Show All Keywords

Keywords/Phrases

Search In:

Publication Years