Multi-stage refinement network for point cloud completion based on geodesic attention
Abstract The attention mechanism has significantly progressed in various point cloud tasks. Benefiting from its significant competence in capturing long-range dependencies, research in point cloud completion has achieved promising results. However, the typically disordered point cloud data features...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-01-01
|
| Series: | Scientific Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1038/s41598-025-86704-6 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832571715657400320 |
|---|---|
| author | Yuchen Chang Kaiping Wang |
| author_facet | Yuchen Chang Kaiping Wang |
| author_sort | Yuchen Chang |
| collection | DOAJ |
| description | Abstract The attention mechanism has significantly progressed in various point cloud tasks. Benefiting from its significant competence in capturing long-range dependencies, research in point cloud completion has achieved promising results. However, the typically disordered point cloud data features complicated non-Euclidean geometric structures and exhibits unpredictable behavior. Most current attention modules are based on Euclidean or local geometry, which fails to accurately represent the intrinsic non-Euclidean characteristics of point cloud data. Thus, we propose a novel geodesic attention-based multi-stage refinement transformer network, which enables the alignment of feature dimensions among query, key, and value, and long-range geometric dependencies are captured on the manifold. Then, a novel Position Feature Extractor is designed to enhance geometric features and explicitly capture graph-based non-Euclidean properties of point cloud objects. A Recurrent Information Aggregation Unit is further applied to aggregate historical information from the previous stages and current geometric features to guide the network in the current stage. The proposed method exhibits strong competitiveness when compared to current state-of-the-art methods. |
| format | Article |
| id | doaj-art-3fb3d78ea1ea4048b45cbdbb6be13802 |
| institution | Kabale University |
| issn | 2045-2322 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj-art-3fb3d78ea1ea4048b45cbdbb6be138022025-02-02T12:21:06ZengNature PortfolioScientific Reports2045-23222025-01-0115111210.1038/s41598-025-86704-6Multi-stage refinement network for point cloud completion based on geodesic attentionYuchen Chang0Kaiping Wang1Department of Computer Science, Xi’an University of Architecture and TechnologyDepartment of Computer Science, Xi’an University of Architecture and TechnologyAbstract The attention mechanism has significantly progressed in various point cloud tasks. Benefiting from its significant competence in capturing long-range dependencies, research in point cloud completion has achieved promising results. However, the typically disordered point cloud data features complicated non-Euclidean geometric structures and exhibits unpredictable behavior. Most current attention modules are based on Euclidean or local geometry, which fails to accurately represent the intrinsic non-Euclidean characteristics of point cloud data. Thus, we propose a novel geodesic attention-based multi-stage refinement transformer network, which enables the alignment of feature dimensions among query, key, and value, and long-range geometric dependencies are captured on the manifold. Then, a novel Position Feature Extractor is designed to enhance geometric features and explicitly capture graph-based non-Euclidean properties of point cloud objects. A Recurrent Information Aggregation Unit is further applied to aggregate historical information from the previous stages and current geometric features to guide the network in the current stage. The proposed method exhibits strong competitiveness when compared to current state-of-the-art methods.https://doi.org/10.1038/s41598-025-86704-6Attention mechanismPoint cloud completionGeodesic attentionRecurrent information aggregation unit |
| spellingShingle | Yuchen Chang Kaiping Wang Multi-stage refinement network for point cloud completion based on geodesic attention Scientific Reports Attention mechanism Point cloud completion Geodesic attention Recurrent information aggregation unit |
| title | Multi-stage refinement network for point cloud completion based on geodesic attention |
| title_full | Multi-stage refinement network for point cloud completion based on geodesic attention |
| title_fullStr | Multi-stage refinement network for point cloud completion based on geodesic attention |
| title_full_unstemmed | Multi-stage refinement network for point cloud completion based on geodesic attention |
| title_short | Multi-stage refinement network for point cloud completion based on geodesic attention |
| title_sort | multi stage refinement network for point cloud completion based on geodesic attention |
| topic | Attention mechanism Point cloud completion Geodesic attention Recurrent information aggregation unit |
| url | https://doi.org/10.1038/s41598-025-86704-6 |
| work_keys_str_mv | AT yuchenchang multistagerefinementnetworkforpointcloudcompletionbasedongeodesicattention AT kaipingwang multistagerefinementnetworkforpointcloudcompletionbasedongeodesicattention |