SAP-Net: A Simple and Robust 3D Point Cloud Registration Network Based on Local Shape Features

Point cloud registration is a key step in the reconstruction of 3D data models. The traditional ICP registration algorithm depends on the initial position of the point cloud. Otherwise, it may get trapped into local optima. In addition, the registration method based on the feature learning of PointN...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Jinlong Li, Yuntao Li, Jiang Long, Yu Zhang, Xiaorong Gao
Formato:	article
Lenguaje:	EN
Publicado:	MDPI AG 2021
Materias:	point cloud registration deep learning feature extraction robustness Chemical technology TP1-1185
Acceso en línea:	https://doaj.org/article/73350980988c4ff89686f359a6a3b0b8
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:73350980988c4ff89686f359a6a3b0b8
record_format	dspace
spelling	oai:doaj.org-article:73350980988c4ff89686f359a6a3b0b82021-11-11T19:10:04ZSAP-Net: A Simple and Robust 3D Point Cloud Registration Network Based on Local Shape Features10.3390/s212171771424-8220https://doaj.org/article/73350980988c4ff89686f359a6a3b0b82021-10-01T00:00:00Zhttps://www.mdpi.com/1424-8220/21/21/7177https://doaj.org/toc/1424-8220Point cloud registration is a key step in the reconstruction of 3D data models. The traditional ICP registration algorithm depends on the initial position of the point cloud. Otherwise, it may get trapped into local optima. In addition, the registration method based on the feature learning of PointNet cannot directly or effectively extract local features. To solve these two problems, this paper proposes SAP-Net, inspired by CorsNet and PointNet++, as an optimized CorsNet. To be more specific, SAP-Net firstly uses the set abstraction layer in PointNet++ as the feature extraction layer and then combines the global features with the initial template point cloud. Finally, PointNet is used as the transform prediction layer to obtain the six parameters required for point cloud registration directly, namely the rotation matrix and the translation vector. Experiments on the ModelNet40 dataset and real data show that SAP-Net not only outperforms ICP and CorsNet on both seen and unseen categories of the point cloud but also has stronger robustness.Jinlong LiYuntao LiJiang LongYu ZhangXiaorong GaoMDPI AGarticlepoint cloudregistrationdeep learningfeature extractionrobustnessChemical technologyTP1-1185ENSensors, Vol 21, Iss 7177, p 7177 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	point cloud registration deep learning feature extraction robustness Chemical technology TP1-1185
spellingShingle	point cloud registration deep learning feature extraction robustness Chemical technology TP1-1185 Jinlong Li Yuntao Li Jiang Long Yu Zhang Xiaorong Gao SAP-Net: A Simple and Robust 3D Point Cloud Registration Network Based on Local Shape Features
description	Point cloud registration is a key step in the reconstruction of 3D data models. The traditional ICP registration algorithm depends on the initial position of the point cloud. Otherwise, it may get trapped into local optima. In addition, the registration method based on the feature learning of PointNet cannot directly or effectively extract local features. To solve these two problems, this paper proposes SAP-Net, inspired by CorsNet and PointNet++, as an optimized CorsNet. To be more specific, SAP-Net firstly uses the set abstraction layer in PointNet++ as the feature extraction layer and then combines the global features with the initial template point cloud. Finally, PointNet is used as the transform prediction layer to obtain the six parameters required for point cloud registration directly, namely the rotation matrix and the translation vector. Experiments on the ModelNet40 dataset and real data show that SAP-Net not only outperforms ICP and CorsNet on both seen and unseen categories of the point cloud but also has stronger robustness.
format	article
author	Jinlong Li Yuntao Li Jiang Long Yu Zhang Xiaorong Gao
author_facet	Jinlong Li Yuntao Li Jiang Long Yu Zhang Xiaorong Gao
author_sort	Jinlong Li
title	SAP-Net: A Simple and Robust 3D Point Cloud Registration Network Based on Local Shape Features
title_short	SAP-Net: A Simple and Robust 3D Point Cloud Registration Network Based on Local Shape Features
title_full	SAP-Net: A Simple and Robust 3D Point Cloud Registration Network Based on Local Shape Features
title_fullStr	SAP-Net: A Simple and Robust 3D Point Cloud Registration Network Based on Local Shape Features
title_full_unstemmed	SAP-Net: A Simple and Robust 3D Point Cloud Registration Network Based on Local Shape Features
title_sort	sap-net: a simple and robust 3d point cloud registration network based on local shape features
publisher	MDPI AG
publishDate	2021
url	https://doaj.org/article/73350980988c4ff89686f359a6a3b0b8
work_keys_str_mv	AT jinlongli sapnetasimpleandrobust3dpointcloudregistrationnetworkbasedonlocalshapefeatures AT yuntaoli sapnetasimpleandrobust3dpointcloudregistrationnetworkbasedonlocalshapefeatures AT jianglong sapnetasimpleandrobust3dpointcloudregistrationnetworkbasedonlocalshapefeatures AT yuzhang sapnetasimpleandrobust3dpointcloudregistrationnetworkbasedonlocalshapefeatures AT xiaoronggao sapnetasimpleandrobust3dpointcloudregistrationnetworkbasedonlocalshapefeatures
_version_	1718431591151173632

SAP-Net: A Simple and Robust 3D Point Cloud Registration Network Based on Local Shape Features

Ejemplares similares