DLAFNet: Direct LiDAR-Aerial Fusion Network for Semantic Segmentation of 2-D Aerial Image and 3-D LiDAR Point Cloud

High-resolution remote sensing image segmentation has advanced significantly with 2-D convolutional neural networks and transformer-based models like SegFormer and Swin Transformer. Concurrently, the rapid development of 3-D convolution techniques has driven advancements in methods like PointNet and...

Full description

Saved in:
Bibliographic Details
Main Authors: Wei Liu, He Wang, Yicheng Qiao, Haopeng Zhang, Junli Yang
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10778434/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846118376714272768
author Wei Liu
He Wang
Yicheng Qiao
Haopeng Zhang
Junli Yang
author_facet Wei Liu
He Wang
Yicheng Qiao
Haopeng Zhang
Junli Yang
author_sort Wei Liu
collection DOAJ
description High-resolution remote sensing image segmentation has advanced significantly with 2-D convolutional neural networks and transformer-based models like SegFormer and Swin Transformer. Concurrently, the rapid development of 3-D convolution techniques has driven advancements in methods like PointNet and Kernel Point Convolution for 3-D LiDAR point cloud segmentation. Traditional fusion of aerial imagery and LiDAR data often relies on digital surface models or other features extracted from LiDAR point clouds, incorporating them as depth channels into image data. In this article, we propose a novel approach called Direct LiDAR-Aerial Fusion Network, which directly integrates multispectral images (RGB) and LiDAR point cloud data for semantic segmentation. Experiments on the modified GRSS18 dataset demonstrate that our method achieves an overall accuracy (OA) of 79.88%, outperforming conventional approaches. By fusing RGB and LiDAR features, our technique improves OA by 1.77% and mean Intersection over Union by 0.83%.
format Article
id doaj-art-cd7c8707f30241ddbb0519a6217e79ac
institution Kabale University
issn 1939-1404
2151-1535
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
spelling doaj-art-cd7c8707f30241ddbb0519a6217e79ac2024-12-18T00:00:29ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-01181864187510.1109/JSTARS.2024.351151710778434DLAFNet: Direct LiDAR-Aerial Fusion Network for Semantic Segmentation of 2-D Aerial Image and 3-D LiDAR Point CloudWei Liu0He Wang1Yicheng Qiao2Haopeng Zhang3https://orcid.org/0000-0003-1981-8307Junli Yang4https://orcid.org/0000-0001-8370-7105School of Artificial Intelligence, Beijing University of Posts and Telecommunications, Beijing, ChinaSino-French Engineer School, Beihang University, Beijing, ChinaSchool of Sports Engineering, Beijing Sport University, Beijing, ChinaSchool of Astronautics, Beihang University, Beijing, ChinaInternational School, Beijing University of Posts and Telecommunications, Beijing, ChinaHigh-resolution remote sensing image segmentation has advanced significantly with 2-D convolutional neural networks and transformer-based models like SegFormer and Swin Transformer. Concurrently, the rapid development of 3-D convolution techniques has driven advancements in methods like PointNet and Kernel Point Convolution for 3-D LiDAR point cloud segmentation. Traditional fusion of aerial imagery and LiDAR data often relies on digital surface models or other features extracted from LiDAR point clouds, incorporating them as depth channels into image data. In this article, we propose a novel approach called Direct LiDAR-Aerial Fusion Network, which directly integrates multispectral images (RGB) and LiDAR point cloud data for semantic segmentation. Experiments on the modified GRSS18 dataset demonstrate that our method achieves an overall accuracy (OA) of 79.88%, outperforming conventional approaches. By fusing RGB and LiDAR features, our technique improves OA by 1.77% and mean Intersection over Union by 0.83%.https://ieeexplore.ieee.org/document/10778434/Aerial imagedata fusionlight detection and ranging (LiDAR)point cloudsemantic segmentation
spellingShingle Wei Liu
He Wang
Yicheng Qiao
Haopeng Zhang
Junli Yang
DLAFNet: Direct LiDAR-Aerial Fusion Network for Semantic Segmentation of 2-D Aerial Image and 3-D LiDAR Point Cloud
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Aerial image
data fusion
light detection and ranging (LiDAR)
point cloud
semantic segmentation
title DLAFNet: Direct LiDAR-Aerial Fusion Network for Semantic Segmentation of 2-D Aerial Image and 3-D LiDAR Point Cloud
title_full DLAFNet: Direct LiDAR-Aerial Fusion Network for Semantic Segmentation of 2-D Aerial Image and 3-D LiDAR Point Cloud
title_fullStr DLAFNet: Direct LiDAR-Aerial Fusion Network for Semantic Segmentation of 2-D Aerial Image and 3-D LiDAR Point Cloud
title_full_unstemmed DLAFNet: Direct LiDAR-Aerial Fusion Network for Semantic Segmentation of 2-D Aerial Image and 3-D LiDAR Point Cloud
title_short DLAFNet: Direct LiDAR-Aerial Fusion Network for Semantic Segmentation of 2-D Aerial Image and 3-D LiDAR Point Cloud
title_sort dlafnet direct lidar aerial fusion network for semantic segmentation of 2 d aerial image and 3 d lidar point cloud
topic Aerial image
data fusion
light detection and ranging (LiDAR)
point cloud
semantic segmentation
url https://ieeexplore.ieee.org/document/10778434/
work_keys_str_mv AT weiliu dlafnetdirectlidaraerialfusionnetworkforsemanticsegmentationof2daerialimageand3dlidarpointcloud
AT hewang dlafnetdirectlidaraerialfusionnetworkforsemanticsegmentationof2daerialimageand3dlidarpointcloud
AT yichengqiao dlafnetdirectlidaraerialfusionnetworkforsemanticsegmentationof2daerialimageand3dlidarpointcloud
AT haopengzhang dlafnetdirectlidaraerialfusionnetworkforsemanticsegmentationof2daerialimageand3dlidarpointcloud
AT junliyang dlafnetdirectlidaraerialfusionnetworkforsemanticsegmentationof2daerialimageand3dlidarpointcloud