Stereollax Net: Stereo Parallax-Based Deep Learning Network for Building Height Estimation DOI
Sana Jabbar, Murtaza Taj

IEEE Transactions on Geoscience and Remote Sensing, Journal Year: 2024, Volume and Issue: 62, P. 1 - 12

Published: Jan. 1, 2024

Accurate estimation of building heights is crucial for effective urban planning and resource management as it provides essential geometric information about the landscape. Many end-to-end deep learning-based networks have been proposed image-to-height mapping using high-resolution non-optical optical remote sensing imagery. In this study, we develop a novel deep-learning architecture that incorporates stereo parallax-based mathematical formulation height estimation. We estimate parameters include differential parallax (ΔP) image, average photo-base (b), satellite (h s ). The final map computed by utilizing these in equation, thus combining closed-form solutions within learning paradigm. Moreover, to improve ΔP, also introduce multi-scale shortcut connections module (MSDSC). MSDSC integrates high-frequency components into lower-resolution baseline decoder features while converting them features. To establish efficacy our network (Stereollax Net), train evaluate method on densely populated cities China (42-Cities dataset) IEEE Data Fusion Contest 2018 dataset (DFC2018). Our Stereollax Net trained only with RGB imagery compared state-of-the-art methods utilize both panchromatic multi-spectral (RGB Near-infrared) qualitative quantitative results demonstrate surpasses existing (SOTA) algorithms, achieving superior performance fewer data training considerable margin. code will be made publicly available via GitHub repository.

Language: Английский

Building Height Extraction From High-Resolution Single-View Remote Sensing Images Using Shadow and Side Information DOI Creative Commons
Wanqi Xu, Zhangyin Feng,

Qian Wan

et al.

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, Journal Year: 2024, Volume and Issue: 17, P. 6514 - 6528

Published: Jan. 1, 2024

Extracting building heights from single-view remote sensing images greatly enhances the application of data. While methods for extracting height shadow have been widely studied, it remains a challenging task. The main reasons are as follows: (1) traditional method information exhibits low accuracy. (2) use only to extract results in limited scenarios. To solve above problems, this paper introduces side and complement each other, proposes extraction high-resolution using information. Firstly, we propose RMU-Net method, which utilizes multi-scale features This aims address issues related pixel detail loss imprecise edge segmentation, result significant scale differences within segmentation targets. Additionally, employ area threshold optimize results, specifically tackle small stray patches holes, enhancing overall integrity accuracy extraction. Secondly, that integrates based on an enhanced proportional coefficient model. measuring lengths is improved by incorporating fishing net informed our analysis geometric relationships among buildings. Finally, establish dataset containing images, select multiple areas experimental analysis. demonstrate 91.03% 90.29%. average absolute error (MAE) 1.22, while root mean square (RMSE) 1.21. Furthermore, proposed method's validity scalability affirmed through analyses applicability anti-interference performance extensive areas.

Language: Английский

Citations

15
Structure-aware deep learning network for building height estimation DOI Creative Commons
Yuehong Chen, Jiayue Zhou, Congcong Xu

et al.

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2025, Volume and Issue: unknown, P. 104443 - 104443

Published: Feb. 1, 2025

Language: Английский

Citations

1
Mapping land- and offshore-based wind turbines in China in 2023 with Sentinel-2 satellite data DOI
Tingting He,

Yihua Hu,

Fashuai Li

et al.

Renewable and Sustainable Energy Reviews, Journal Year: 2025, Volume and Issue: 214, P. 115566 - 115566

Published: March 1, 2025

Language: Английский

Citations

1
Global spatial patterns between nighttime light intensity and urban building morphology DOI Creative Commons
Bin Wu, Hailan Huang, Yu Wang

et al.

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2023, Volume and Issue: 124, P. 103495 - 103495

Published: Sept. 20, 2023

The comprehensive characterization of global urbanization requires consideration both human activities and urban physical structures. Both structures exhibit regular self-similar patterns, yet the spatial patterns between two at a scale remain elusive. This study utilized NPP-VIIRS annual composite data newly available world settlement footprint 3D to investigate relationships nighttime light intensity building morphological indicators across several scales. Our results demonstrated that there is weak association morphology pixel level, as shown by maximum correlation coefficient approximately 0.4, but strong provincial/state level with over 0.8. Additionally, we performed an urban-rural gradient analysis evaluate indicators. indicated dominant gradients for morphologies follow declining trend from centers rural areas. Notably, inconsistencies were found predominantly in Africa. findings also suggested can be served indicator urbanization, thus provide implications facilitating solutions aimed reducing income disparity promoting sustainable development.

Language: Английский

Citations

18
Reconstruction of 3D Information of Buildings from Single-View Images Based on Shadow Information DOI Creative Commons
Zhixin Li, Song Ji, Dazhao Fan

et al.

ISPRS International Journal of Geo-Information, Journal Year: 2024, Volume and Issue: 13(3), P. 62 - 62

Published: Feb. 20, 2024

Accurate building geometry information is crucial for urban planning in constrained spaces, fueling the growing demand large-scale, high-precision 3D city modeling. Traditional methods like oblique photogrammetry and LiDAR prove time consuming expensive low-cost reconstruction of expansive scenes. Addressing this challenge, our study proposes a novel approach to leveraging single-view remote sensing images. By integrating shadow with deep learning networks, method measures height employs semantic segmentation technique single-image high-rise reconstruction. In addition, we have designed complex measurement algorithms contour correction improve accuracy models conjunction previous research. We evaluate method’s precision, efficiency, applicability across various data sources, scenarios, scales. The results demonstrate rapid accurate acquisition maintained geometric (mean error below 5 m). This offers an economical effective solution large-scale modeling, bridging gap cost-efficient techniques.

Language: Английский

Citations

6
Urban building height extraction accommodating various terrain scenes using ICESat-2/ATLAS data DOI Creative Commons
Xiang Huang, Feng Cheng, Yinli Bao

et al.

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2024, Volume and Issue: 130, P. 103870 - 103870

Published: May 15, 2024

Although the photon point cloud data acquired from ICESat-2/ATLAS can be efficiently employed in urban building height extraction, its universal applicability undulating terrain scenarios is constrained, and there are noticeable issues of false positives negatives. This research establishes a terrain-adaptive methodological framework based on to extract high-precision, high-density across varied topographical conditions. First, elevation buffer utilized coarse denoise cloud, involving removal majority noise photons scene, thereby enhancing efficiency subsequent algorithms. Second, signal extracted remaining original using Adaptive Method Based Single-Photon Spatial Distribution (SPSD-AM). approach demonstrates high universality various scenes, while simultaneously ensuring stable accuracy extraction. Subsequently, ground fit curve Differences Urban Signal Photons (USPSD-AM), which addresses challenge potential mixing complex scenarios. A precise then photons. In order mitigate such as negatives, post-processing steps, including completion denoising photons, implemented. Finally, adopted accurate parameters. The precision verification results show that heights considerably consistent with reference heights. mean RMSE MAE 0.273 m 0.202 for flat terrains 1.168 0.759 terrains, respectively. proposed method superior diverse scenarios, providing robust theoretical foundation large-scale retrieval efforts.

Language: Английский

Citations

5
Utilizing Building Offset and Shadow to Retrieve Urban Building Heights with ICESat-2 Photons DOI Creative Commons
Bin Wu, Hailan Huang, Yi Zhao

et al.

Remote Sensing, Journal Year: 2023, Volume and Issue: 15(15), P. 3786 - 3786

Published: July 30, 2023

Building height serves as an essential feature of urban morphology that provides valuable insights into human socio-cultural behaviors and their impact on the environment in milieu. However, openly accessible building information at individual level is still lacking remains sorely limited. Previous studies have shown ICESat-2′s ATL03/08 products are good accuracy for heights retrieval, however, these limited to areas with available data coverage. To this end, we propose a method extracting by using ICESat-2 ATL03 photons high-resolution remote sensing images. We first extracted roof footprint offsets shadows from high resolution imagery multitasking CNN frameworks. Using samples calculated photons, developed estimation combines offset shadow length information. assessed efficacy proposed Wujiaochang area Shanghai city, China. The results indicated able extract MAE 4.7 m, outperforms traditional shadow-based offset-based method. believe candidate accurately retrieving city-wide scale.

Language: Английский

Citations

12
Automated Building Height Estimation Using Ice, Cloud, and Land Elevation Satellite 2 Light Detection and Ranging Data and Building Footprints DOI Creative Commons

Panli Cai,

Jingxian Guo,

Runkui Li

et al.

Remote Sensing, Journal Year: 2024, Volume and Issue: 16(2), P. 263 - 263

Published: Jan. 9, 2024

Accurately estimating building heights is crucial for various applications, including urban planning, climate studies, population estimation, and environmental assessment. However, this remains a challenging task, particularly large areas. Satellite-based Light Detection Ranging (LiDAR) has shown promise, but it often faces difficulties in distinguishing photons from other ground objects. To address challenge, we propose novel method that incorporates footprints, relative positions of photons, self-adaptive buffer photon selection. We employ the Ice, Cloud, Land Elevation Satellite 2 (ICESat-2) photon-counting LiDAR, specifically ICESat-2/ATL03 data, along with footprints obtained New York City (NYC) Open Data platform. The proposed approach was applied to estimate 17,399 buildings NYC, results showed strong consistency reference heights. root mean square error (RMSE) 8.1 m, 71% buildings, absolute (MAE) less than 3 m. Furthermore, conducted an extensive evaluation thoroughly investigated influence terrain, region, height, density, parameter also verified effectiveness our experimental area Beijing compared existing methods. By leveraging ICESat-2 LiDAR advanced selection techniques, demonstrates potential accurately over broad

Language: Английский

Citations

5
Fusing multimodal data of nature-economy-society for large-scale urban building height estimation DOI Creative Commons
Shouhang Du, Hao Liu, Jianghe Xing

et al.

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2024, Volume and Issue: 129, P. 103809 - 103809

Published: April 5, 2024

The building height holds significant importance for comprehensively understanding urban morphology, enhancing planning, and fostering sustainable development. Although many methods using optical SAR images have been presented estimation, these fall short in capturing the influences of economic social attributes on height. In this study, we introduced a Nature-Economy-Society (NES) feature model to represent information, established multi-scale One-Dimensional (1-D) Convolutional Neural Network predicting heights, referred as NES-CNN. First, derived natural buildings from time-series Sentinel-1 Sentinel-2 multispectral images, well World Settlement Footprint (WSF) data Digital Elevation Model (DEM), nighttime light Gross Domestic Product (GDP) data, function Points Interest (POI) data. Second, an autoencoder is employed reduce dimensionality high-dimensional attribute features, minimizing redundancy. Finally, 1-D CNN explore correlations between multi-source heterogeneous NES features facilitating prediction experiments, applied proposed method estimate heights Beijing Shanghai at spatial resolution 10 m. results indicated that Beijing, RMSE, MAE, R values are 6.93 m, 4.41 0.84, respectively, while Shanghai, 7.57 5.38 0.80, respectively. addition information decreases RMSE by 6 % both compared with only attributes. comparison existing studies same mapping resolution, 39 51 Shanghai. innovative inspiring nature study lies its application large-scale estimation.

Language: Английский

Citations

4
Generating high-resolution DEMs in mountainous regions using ICESat-2/ATLAS photons DOI
Yi Zhao, Bin Wu, Gefei Kong

et al.

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2025, Volume and Issue: 138, P. 104461 - 104461

Published: March 14, 2025

Language: Английский

Citations

0