Multi-Feature Aggregation for Semantic Segmentation of an Urban Scene Point Cloud
With the rapid development of cities, semantic segmentation of urban scenes, as an important and effective imaging method, can accurately obtain the distribution information of typical urban ground features, reflecting the development scale and the level of greenery in the cities. There are some cha...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-10-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/14/20/5134 |
| _version_ | 1827648082042421248 |
|---|---|
| author | Jiaqing Chen Yindi Zhao Congtang Meng Yang Liu |
| author_facet | Jiaqing Chen Yindi Zhao Congtang Meng Yang Liu |
| author_sort | Jiaqing Chen |
| collection | DOAJ |
| description | With the rapid development of cities, semantic segmentation of urban scenes, as an important and effective imaging method, can accurately obtain the distribution information of typical urban ground features, reflecting the development scale and the level of greenery in the cities. There are some challenging problems in the semantic segmentation of point clouds in urban scenes, including different scales, imbalanced class distribution, and missing data caused by occlusion. Based on the point cloud semantic segmentation network RandLA-Net, we propose the semantic segmentation networks RandLA-Net++ and RandLA-Net3+. The RandLA-Net++ network is a deep fusion of the shallow and deep features of the point clouds, and a series of nested dense skip connections is used between the encoder and decoder. RandLA-Net3+ is based on the multi-scale connection between the encoder and decoder; it also connects internally within the decoder to capture fine-grained details and coarse-grained semantic information at a full scale. We also propose incorporating dilated convolution to increase the receptive field and compare the improvement effect of different loss functions on sample class imbalance. After verification and analysis of our labeled urban scene LiDAR point cloud dataset—called NJSeg-3D—the mIoU of the RandLA-Net++ and RandLA-Net3+ networks is 3.4% and 3.2% higher, respectively, than the benchmark network RandLA-Net. |
| first_indexed | 2024-03-09T19:31:27Z |
| format | Article |
| id | doaj.art-2a6be52927994280af92e0b74118e408 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-09T19:31:27Z |
| publishDate | 2022-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-2a6be52927994280af92e0b74118e4082023-11-24T02:19:50ZengMDPI AGRemote Sensing2072-42922022-10-011420513410.3390/rs14205134Multi-Feature Aggregation for Semantic Segmentation of an Urban Scene Point CloudJiaqing Chen0Yindi Zhao1Congtang Meng2Yang Liu3Key Laboratory of Degraded and Unused Land Consolidation Engineering, Ministry of Natural Resources, Xi’an 710075, ChinaKey Laboratory of Degraded and Unused Land Consolidation Engineering, Ministry of Natural Resources, Xi’an 710075, ChinaSchool of Environment and Spatial Informatics, China University of Mining and Technology, Xuzhou 221116, ChinaSchool of Environment and Spatial Informatics, China University of Mining and Technology, Xuzhou 221116, ChinaWith the rapid development of cities, semantic segmentation of urban scenes, as an important and effective imaging method, can accurately obtain the distribution information of typical urban ground features, reflecting the development scale and the level of greenery in the cities. There are some challenging problems in the semantic segmentation of point clouds in urban scenes, including different scales, imbalanced class distribution, and missing data caused by occlusion. Based on the point cloud semantic segmentation network RandLA-Net, we propose the semantic segmentation networks RandLA-Net++ and RandLA-Net3+. The RandLA-Net++ network is a deep fusion of the shallow and deep features of the point clouds, and a series of nested dense skip connections is used between the encoder and decoder. RandLA-Net3+ is based on the multi-scale connection between the encoder and decoder; it also connects internally within the decoder to capture fine-grained details and coarse-grained semantic information at a full scale. We also propose incorporating dilated convolution to increase the receptive field and compare the improvement effect of different loss functions on sample class imbalance. After verification and analysis of our labeled urban scene LiDAR point cloud dataset—called NJSeg-3D—the mIoU of the RandLA-Net++ and RandLA-Net3+ networks is 3.4% and 3.2% higher, respectively, than the benchmark network RandLA-Net.https://www.mdpi.com/2072-4292/14/20/5134point cloud semantic segmentationmulti-feature aggregationRandLA-Netdeep learning |
| spellingShingle | Jiaqing Chen Yindi Zhao Congtang Meng Yang Liu Multi-Feature Aggregation for Semantic Segmentation of an Urban Scene Point Cloud Remote Sensing point cloud semantic segmentation multi-feature aggregation RandLA-Net deep learning |
| title | Multi-Feature Aggregation for Semantic Segmentation of an Urban Scene Point Cloud |
| title_full | Multi-Feature Aggregation for Semantic Segmentation of an Urban Scene Point Cloud |
| title_fullStr | Multi-Feature Aggregation for Semantic Segmentation of an Urban Scene Point Cloud |
| title_full_unstemmed | Multi-Feature Aggregation for Semantic Segmentation of an Urban Scene Point Cloud |
| title_short | Multi-Feature Aggregation for Semantic Segmentation of an Urban Scene Point Cloud |
| title_sort | multi feature aggregation for semantic segmentation of an urban scene point cloud |
| topic | point cloud semantic segmentation multi-feature aggregation RandLA-Net deep learning |
| url | https://www.mdpi.com/2072-4292/14/20/5134 |
| work_keys_str_mv | AT jiaqingchen multifeatureaggregationforsemanticsegmentationofanurbanscenepointcloud AT yindizhao multifeatureaggregationforsemanticsegmentationofanurbanscenepointcloud AT congtangmeng multifeatureaggregationforsemanticsegmentationofanurbanscenepointcloud AT yangliu multifeatureaggregationforsemanticsegmentationofanurbanscenepointcloud |