Open-source data-driven urban land-use mapping integrating point-line-polygon semantic objects: A case study of Chinese cities

Zhong, Yanfei; Su, Yu; Wu, Siqi; Zheng, Zhendong; Zhao, Ji; Ma, Ailong; Zhu, Qiqi; Ye, Richen; Li, Xiaoman; Pellikka, Petri; Zhang, Liangpei

doi:10.1016/j.rse.2020.111838

Cited by 80 publications

(31 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several studies have used OSM and remote sensing for urban mapping (see Table 1). As far as we can see, OSM was used to define samples (Di Yang et al, 2017;Liu et al, 2020 andZhang, Gorelick andZimba, 2020), to determine land parceling (CHEN, W. et al, 2018;CHEN, W. et al, 2018;SUN, J. et al, 2020;Zhong et al 2020 andCHEN, et al, 2021) and to assist in obtaining semantic objects (Zhao et al, 2019 andZhong et al, 2020), which were subsequently processed by artificial-intelligence algorithms for producing mappings, which delimit distinct classes. This research demonstrated the possibility of using data from OSM in conjunction with orbital images to map urban and non-urban features involving extensive territorial areas, using relatively simple but robust techniques in terms of practical results.…”

Section: Discussionmentioning

confidence: 99%

Proposal for an index of roads and structures for the mapping of non-vegetated urban surfaces using OSM and Sentinel data

Justiniano¹,

Junior²,

Melo³

et al. 2022

Preprint

View full text Add to dashboard Cite

The use of volunteered geographic information (VGI), such as OpenStreetMap (OSM), to assist in mapping land use and coverage together with remote sensing images is relatively recent. Most studies have used OSM to assist in sample collection for image classification or aggregated vectors of buildings, transport, and land-use and coverage as ancillary data to support mapping refinements. This study proposes a metric called the “index of roads and structures” (IRS) created on the basis of OSM data with the intention of assisting in the mapping of non-vegetated and non-aquatic urban surfaces. IRS thresholds were defined and supplemented with information derived from the Normalized Difference Vegetation Index (NDVI) and the Modified Normalized Difference Water Index (MNDWI) as a way of extending the restriction between urban and non-urban classes and thus achieving better mapping accuracy. To implement this study, multispectral Sentinel 2 images resampled to 10 m on the ground were processed in the Google Earth Engine (GEE). The IRS is a raster file, in which each pixel is associated with the possibility of being inserted in an urban context; thus, the importance of this index as an aid in mapping urban areas is clear. We have demonstrated the possibility of using the IRS to map urban surfaces in Brazil (8.5 million square kilometers) and obtained a very high accuracy of 93.2%.

show abstract

Section: Discussionmentioning

confidence: 99%

Proposal for an index of roads and structures for the mapping of non-vegetated urban surfaces using OSM and Sentinel data

Justiniano¹,

Junior²,

Melo³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Besides, we aim to thoroughly evaluate the performance of the proposed MTDL approach in the scene-level 3D building mapping. Although DL models have set new benchmarks on several remote-sensing image processing tasks (land use classification Zhong et al, 2020, building segmentation Shi et al, 2020, local climate zone classification Zhu et al, 2022, etc. ), whether DL methods outperform ML ones on the scene-level 3D building mapping is yet to be examined.…”

Section: Have Two Major Limitationsmentioning

confidence: 99%

SHAFTS (v2022.3): a deep-learning-based Python package for simultaneous extraction of building height and footprint from sentinel imagery

Sun

2023

Geosci. Model Dev.

View full text Add to dashboard Cite

Abstract. Building height and footprint are two fundamental urban morphological features required by urban climate modelling. Although some statistical methods have been proposed to estimate average building height and footprint from publicly available satellite imagery, they often involve tedious feature engineering which makes it hard to achieve efficient knowledge discovery in a changing urban environment with ever-increasing earth observations. In this work, we develop a deep-learning-based (DL) Python package – SHAFTS (Simultaneous building Height And FootprinT extraction from Sentinel imagery) to extract such information. Multi-task deep-learning (MTDL) models are proposed to automatically learn feature representation shared by building height and footprint prediction. Besides, we integrate digital elevation model (DEM) information into developed models to inform models of terrain-induced effects on the backscattering displayed by Sentinel-1 imagery. We set conventional machine-learning-based (ML) models and single-task deep-learning (STDL) models as benchmarks and select 46 cities worldwide to evaluate developed models’ patch-level prediction skills and city-level spatial transferability at four resolutions (100, 250, 500 and 1000 m). Patch-level results of 43 cities show that DL models successfully produce discriminative feature representation and improve the coefficient of determination (R2) of building height and footprint prediction more than ML models by 0.27–0.63 and 0.11–0.49, respectively. Moreover, stratified error assessment reveals that DL models effectively mitigate the severe systematic underestimation of ML models in the high-value domain: for the 100 m case, DL models reduce the root mean square error (RMSE) of building height higher than 40 m and building footprint larger than 0.25 by 31 m and 0.1, respectively, which demonstrates the superiority of DL models on refined 3D building information extraction in highly urbanized areas. For the evaluation of spatial transferability, when compared with an existing state-of-the-art product, DL models can achieve similar improvement on the overall performance and high-value prediction. Furthermore, within the DL family, comparison in building height prediction between STDL and MTDL models reveals that MTDL models achieve higher accuracy in all cases and smaller bias uncertainty for the prediction in the high-value domain at the refined scale, which proves the effectiveness of multi-task learning (MTL) on building height estimation.

show abstract

“…Deep Convolutional Neural Networks (DCNNs) are able to adaptively approximate the relationship between image information and land information through multi-layer transformations ( Zhu et al, 2017 ). Thus, compared with conventional land cover classification methods, deep models can accurately characterize complex contextual information contained in high-resolution images ( Tong et al, 2020 , Huang et al, 2018a , Zhang et al, 2019 , Srivastava et al, 2019 , Zhong et al, 2020 ). Although deep models have reported great superiorities in many remote sensing issues ( Zhu et al, 2017 , Ma et al, 2019 , Zhu et al, 2021 ), their performance strongly relies on the quality and quantity of training data ( LeCun et al, 2015 , Xia et al, 2017 , Ding et al, 2021 ), resulting in two main problems in applying them to real-world land cover mapping:…”

Section: Introductionmentioning

confidence: 99%

Enabling country-scale land cover mapping with meter-resolution satellite imagery

Tong¹,

Xia²,

Zhu³

2023

ISPRS Journal of Photogrammetry and Remote Sensing

View full text Add to dashboard Cite

Open-source data-driven urban land-use mapping integrating point-line-polygon semantic objects: A case study of Chinese cities

Cited by 80 publications

References 27 publications

Proposal for an index of roads and structures for the mapping of non-vegetated urban surfaces using OSM and Sentinel data

Proposal for an index of roads and structures for the mapping of non-vegetated urban surfaces using OSM and Sentinel data

SHAFTS (v2022.3): a deep-learning-based Python package for simultaneous extraction of building height and footprint from sentinel imagery

Enabling country-scale land cover mapping with meter-resolution satellite imagery

Contact Info

Product

Resources

About