Flattening-net: deep regular 2D representation for 3D point cloud analysis
Point clouds are characterized by irregularity and unstructuredness, which pose challenges in efficient data exploitation and discriminative feature extraction. In this paper, we present an unsupervised deep neural architecture called Flattening-Net to represent irregular 3D point clouds of arbitrar...
Saved in:
Main Authors: | , , , , , |
---|---|
Other Authors: | |
Format: | Article |
Language: | English |
Published: |
2023
|
Subjects: | |
Online Access: | https://hdl.handle.net/10356/172180 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Nanyang Technological University |
Language: | English |
id |
sg-ntu-dr.10356-172180 |
---|---|
record_format |
dspace |
spelling |
sg-ntu-dr.10356-1721802023-11-28T06:04:49Z Flattening-net: deep regular 2D representation for 3D point cloud analysis Zhang, Qijian Hou, Junhui Qian, Yue Zeng, Yiming Zhang, Juyong He, Ying School of Computer Science and Engineering Engineering::Computer science and engineering Deep Neural Network Point Cloud Point clouds are characterized by irregularity and unstructuredness, which pose challenges in efficient data exploitation and discriminative feature extraction. In this paper, we present an unsupervised deep neural architecture called Flattening-Net to represent irregular 3D point clouds of arbitrary geometry and topology as a completely regular 2D point geometry image (PGI) structure, in which coordinates of spatial points are captured in colors of image pixels. Intuitively, Flattening-Net implicitly approximates a locally smooth 3D-to-2D surface flattening process while effectively preserving neighborhood consistency. As a generic representation modality, PGI inherently encodes the intrinsic property of the underlying manifold structure and facilitates surface-style point feature aggregation. To demonstrate its potential, we construct a unified learning framework directly operating on PGIs to achieve diverse types of high-level and low-level downstream applications driven by specific task networks, including classification, segmentation, reconstruction, and upsampling. Extensive experiments demonstrate that our methods perform favorably against the current state-of-the-art competitors. This work was supported in part by the Hong Kong Research Grants Council under Grants 11219422, 11202320, and 11218121, and in part by the Natural Science Foundation of China under Grant 61871342, in part by Hong Kong Innovation and Technology Fund under Grant MHP/117/21, and in part by the Basic Research General Program of Shenzhen Municipality under Grant JCYJ20190808183003968. 2023-11-28T06:04:49Z 2023-11-28T06:04:49Z 2023 Journal Article Zhang, Q., Hou, J., Qian, Y., Zeng, Y., Zhang, J. & He, Y. (2023). Flattening-net: deep regular 2D representation for 3D point cloud analysis. IEEE Transactions On Pattern Analysis and Machine Intelligence, 45(8), 9726-9742. https://dx.doi.org/10.1109/TPAMI.2023.3244828 0162-8828 https://hdl.handle.net/10356/172180 10.1109/TPAMI.2023.3244828 37022866 2-s2.0-85149363623 8 45 9726 9742 en IEEE Transactions on Pattern Analysis and Machine Intelligence © 2023 IEEE. All rights reserved. |
institution |
Nanyang Technological University |
building |
NTU Library |
continent |
Asia |
country |
Singapore Singapore |
content_provider |
NTU Library |
collection |
DR-NTU |
language |
English |
topic |
Engineering::Computer science and engineering Deep Neural Network Point Cloud |
spellingShingle |
Engineering::Computer science and engineering Deep Neural Network Point Cloud Zhang, Qijian Hou, Junhui Qian, Yue Zeng, Yiming Zhang, Juyong He, Ying Flattening-net: deep regular 2D representation for 3D point cloud analysis |
description |
Point clouds are characterized by irregularity and unstructuredness, which pose challenges in efficient data exploitation and discriminative feature extraction. In this paper, we present an unsupervised deep neural architecture called Flattening-Net to represent irregular 3D point clouds of arbitrary geometry and topology as a completely regular 2D point geometry image (PGI) structure, in which coordinates of spatial points are captured in colors of image pixels. Intuitively, Flattening-Net implicitly approximates a locally smooth 3D-to-2D surface flattening process while effectively preserving neighborhood consistency. As a generic representation modality, PGI inherently encodes the intrinsic property of the underlying manifold structure and facilitates surface-style point feature aggregation. To demonstrate its potential, we construct a unified learning framework directly operating on PGIs to achieve diverse types of high-level and low-level downstream applications driven by specific task networks, including classification, segmentation, reconstruction, and upsampling. Extensive experiments demonstrate that our methods perform favorably against the current state-of-the-art competitors. |
author2 |
School of Computer Science and Engineering |
author_facet |
School of Computer Science and Engineering Zhang, Qijian Hou, Junhui Qian, Yue Zeng, Yiming Zhang, Juyong He, Ying |
format |
Article |
author |
Zhang, Qijian Hou, Junhui Qian, Yue Zeng, Yiming Zhang, Juyong He, Ying |
author_sort |
Zhang, Qijian |
title |
Flattening-net: deep regular 2D representation for 3D point cloud analysis |
title_short |
Flattening-net: deep regular 2D representation for 3D point cloud analysis |
title_full |
Flattening-net: deep regular 2D representation for 3D point cloud analysis |
title_fullStr |
Flattening-net: deep regular 2D representation for 3D point cloud analysis |
title_full_unstemmed |
Flattening-net: deep regular 2D representation for 3D point cloud analysis |
title_sort |
flattening-net: deep regular 2d representation for 3d point cloud analysis |
publishDate |
2023 |
url |
https://hdl.handle.net/10356/172180 |
_version_ |
1783955554488549376 |