PU-CTG: A Point Cloud Upsampling Network Using Transformer Fusion and GRU Correction
Point clouds are widely used in remote sensing applications, e.g., 3D object classification, semantic segmentation, and building reconstruction. Generating dense and uniformly distributed point clouds from low-density ones is beneficial to 3D point cloud applications. The traditional methods mainly...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-01-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/16/3/450 |
| _version_ | 1797318276690214912 |
|---|---|
| author | Tianyu Li Yanghong Lin Bo Cheng Guo Ai Jian Yang Li Fang |
| author_facet | Tianyu Li Yanghong Lin Bo Cheng Guo Ai Jian Yang Li Fang |
| author_sort | Tianyu Li |
| collection | DOAJ |
| description | Point clouds are widely used in remote sensing applications, e.g., 3D object classification, semantic segmentation, and building reconstruction. Generating dense and uniformly distributed point clouds from low-density ones is beneficial to 3D point cloud applications. The traditional methods mainly focus on the global shape of 3D point clouds, thus ignoring detailed representations. The enhancement of detailed features is conducive to generating dense and uniform point clouds. In this paper, we propose a point cloud upsampling network to improve the detail construction ability, named PU-CTG. The proposed method is implemented by a cross-transformer-fused module and a GRU-corrected module. The aim of the cross-transformer module is to enable the interaction and effective fusion between different scales of features so that the network can capture finer features. The purpose of the gated recurrent unit (GRU) is to reconstruct fine-grained features by rectifying the feedback error. The experimental results demonstrate the effectiveness of our method. Furthermore, the ModelNet40 dataset is upsampled by PU-CTG, and the classification experiment is applied to PointNet to verify the promotion ability of this network. |
| first_indexed | 2024-03-08T03:50:07Z |
| format | Article |
| id | doaj.art-b4ac5087a01449439241370590586b4c |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-08T03:50:07Z |
| publishDate | 2024-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-b4ac5087a01449439241370590586b4c2024-02-09T15:21:08ZengMDPI AGRemote Sensing2072-42922024-01-0116345010.3390/rs16030450PU-CTG: A Point Cloud Upsampling Network Using Transformer Fusion and GRU CorrectionTianyu Li0Yanghong Lin1Bo Cheng2Guo Ai3Jian Yang4Li Fang5Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou 350117, ChinaFujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou 350117, ChinaSchool of Computer and Cyberspace Security, Fujian Normal University, Fuzhou 350117, ChinaQuanzhou Institute of Equipment Manufacturing Haixi Institutes, Chinese Academy of Sciences, Quanzhou 362216, ChinaSchool of Geospatial Information, Information Engineering University, Zhengzhou 450052, ChinaQuanzhou Institute of Equipment Manufacturing Haixi Institutes, Chinese Academy of Sciences, Quanzhou 362216, ChinaPoint clouds are widely used in remote sensing applications, e.g., 3D object classification, semantic segmentation, and building reconstruction. Generating dense and uniformly distributed point clouds from low-density ones is beneficial to 3D point cloud applications. The traditional methods mainly focus on the global shape of 3D point clouds, thus ignoring detailed representations. The enhancement of detailed features is conducive to generating dense and uniform point clouds. In this paper, we propose a point cloud upsampling network to improve the detail construction ability, named PU-CTG. The proposed method is implemented by a cross-transformer-fused module and a GRU-corrected module. The aim of the cross-transformer module is to enable the interaction and effective fusion between different scales of features so that the network can capture finer features. The purpose of the gated recurrent unit (GRU) is to reconstruct fine-grained features by rectifying the feedback error. The experimental results demonstrate the effectiveness of our method. Furthermore, the ModelNet40 dataset is upsampled by PU-CTG, and the classification experiment is applied to PointNet to verify the promotion ability of this network.https://www.mdpi.com/2072-4292/16/3/450point cloud upsamplinggenerative adversarial networkscross-transformergated recurrent unitpoint cloud classification |
| spellingShingle | Tianyu Li Yanghong Lin Bo Cheng Guo Ai Jian Yang Li Fang PU-CTG: A Point Cloud Upsampling Network Using Transformer Fusion and GRU Correction Remote Sensing point cloud upsampling generative adversarial networks cross-transformer gated recurrent unit point cloud classification |
| title | PU-CTG: A Point Cloud Upsampling Network Using Transformer Fusion and GRU Correction |
| title_full | PU-CTG: A Point Cloud Upsampling Network Using Transformer Fusion and GRU Correction |
| title_fullStr | PU-CTG: A Point Cloud Upsampling Network Using Transformer Fusion and GRU Correction |
| title_full_unstemmed | PU-CTG: A Point Cloud Upsampling Network Using Transformer Fusion and GRU Correction |
| title_short | PU-CTG: A Point Cloud Upsampling Network Using Transformer Fusion and GRU Correction |
| title_sort | pu ctg a point cloud upsampling network using transformer fusion and gru correction |
| topic | point cloud upsampling generative adversarial networks cross-transformer gated recurrent unit point cloud classification |
| url | https://www.mdpi.com/2072-4292/16/3/450 |
| work_keys_str_mv | AT tianyuli puctgapointcloudupsamplingnetworkusingtransformerfusionandgrucorrection AT yanghonglin puctgapointcloudupsamplingnetworkusingtransformerfusionandgrucorrection AT bocheng puctgapointcloudupsamplingnetworkusingtransformerfusionandgrucorrection AT guoai puctgapointcloudupsamplingnetworkusingtransformerfusionandgrucorrection AT jianyang puctgapointcloudupsamplingnetworkusingtransformerfusionandgrucorrection AT lifang puctgapointcloudupsamplingnetworkusingtransformerfusionandgrucorrection |