Building Polygon Extraction from Aerial Images and Digital Surface Models with a Frame Field Learning Framework

Deep learning-based models for building delineation from remotely sensed images face the challenge of producing precise and regular building outlines. This study investigates the combination of normalized digital surface models (nDSMs) with aerial images to optimize the extraction of building polygo...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Xiaoyu Sun, Wufan Zhao, Raian V. Maretto, Claudio Persello
Formato:	article
Lenguaje:	EN
Publicado:	MDPI AG 2021
Materias:	building outline delineation convolutional neural networks regularized polygonization frame field Science Q
Acceso en línea:	https://doaj.org/article/70adfefe4fae4471ae85672fb902b7db
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:70adfefe4fae4471ae85672fb902b7db
record_format	dspace
spelling	oai:doaj.org-article:70adfefe4fae4471ae85672fb902b7db2021-11-25T18:55:32ZBuilding Polygon Extraction from Aerial Images and Digital Surface Models with a Frame Field Learning Framework10.3390/rs132247002072-4292https://doaj.org/article/70adfefe4fae4471ae85672fb902b7db2021-11-01T00:00:00Zhttps://www.mdpi.com/2072-4292/13/22/4700https://doaj.org/toc/2072-4292Deep learning-based models for building delineation from remotely sensed images face the challenge of producing precise and regular building outlines. This study investigates the combination of normalized digital surface models (nDSMs) with aerial images to optimize the extraction of building polygons using the frame field learning method. Results are evaluated at pixel, object, and polygon levels. In addition, an analysis is performed to assess the statistical deviations in the number of vertices of building polygons compared with the reference. The comparison of the number of vertices focuses on finding the output polygons that are the easiest to edit by human analysts in operational applications. It can serve as guidance to reduce the post-processing workload for obtaining high-accuracy building footprints. Experiments conducted in Enschede, the Netherlands, demonstrate that by introducing nDSM, the method could reduce the number of false positives and prevent missing the real buildings on the ground. The positional accuracy and shape similarity was improved, resulting in better-aligned building polygons. The method achieved a mean intersection over union (IoU) of 0.80 with the fused data (RGB + nDSM) against an IoU of 0.57 with the baseline (using RGB only) in the same area. A qualitative analysis of the results shows that the investigated model predicts more precise and regular polygons for large and complex structures.Xiaoyu SunWufan ZhaoRaian V. MarettoClaudio PerselloMDPI AGarticlebuilding outline delineationconvolutional neural networksregularized polygonizationframe fieldScienceQENRemote Sensing, Vol 13, Iss 4700, p 4700 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	building outline delineation convolutional neural networks regularized polygonization frame field Science Q
spellingShingle	building outline delineation convolutional neural networks regularized polygonization frame field Science Q Xiaoyu Sun Wufan Zhao Raian V. Maretto Claudio Persello Building Polygon Extraction from Aerial Images and Digital Surface Models with a Frame Field Learning Framework
description	Deep learning-based models for building delineation from remotely sensed images face the challenge of producing precise and regular building outlines. This study investigates the combination of normalized digital surface models (nDSMs) with aerial images to optimize the extraction of building polygons using the frame field learning method. Results are evaluated at pixel, object, and polygon levels. In addition, an analysis is performed to assess the statistical deviations in the number of vertices of building polygons compared with the reference. The comparison of the number of vertices focuses on finding the output polygons that are the easiest to edit by human analysts in operational applications. It can serve as guidance to reduce the post-processing workload for obtaining high-accuracy building footprints. Experiments conducted in Enschede, the Netherlands, demonstrate that by introducing nDSM, the method could reduce the number of false positives and prevent missing the real buildings on the ground. The positional accuracy and shape similarity was improved, resulting in better-aligned building polygons. The method achieved a mean intersection over union (IoU) of 0.80 with the fused data (RGB + nDSM) against an IoU of 0.57 with the baseline (using RGB only) in the same area. A qualitative analysis of the results shows that the investigated model predicts more precise and regular polygons for large and complex structures.
format	article
author	Xiaoyu Sun Wufan Zhao Raian V. Maretto Claudio Persello
author_facet	Xiaoyu Sun Wufan Zhao Raian V. Maretto Claudio Persello
author_sort	Xiaoyu Sun
title	Building Polygon Extraction from Aerial Images and Digital Surface Models with a Frame Field Learning Framework
title_short	Building Polygon Extraction from Aerial Images and Digital Surface Models with a Frame Field Learning Framework
title_full	Building Polygon Extraction from Aerial Images and Digital Surface Models with a Frame Field Learning Framework
title_fullStr	Building Polygon Extraction from Aerial Images and Digital Surface Models with a Frame Field Learning Framework
title_full_unstemmed	Building Polygon Extraction from Aerial Images and Digital Surface Models with a Frame Field Learning Framework
title_sort	building polygon extraction from aerial images and digital surface models with a frame field learning framework
publisher	MDPI AG
publishDate	2021
url	https://doaj.org/article/70adfefe4fae4471ae85672fb902b7db
work_keys_str_mv	AT xiaoyusun buildingpolygonextractionfromaerialimagesanddigitalsurfacemodelswithaframefieldlearningframework AT wufanzhao buildingpolygonextractionfromaerialimagesanddigitalsurfacemodelswithaframefieldlearningframework AT raianvmaretto buildingpolygonextractionfromaerialimagesanddigitalsurfacemodelswithaframefieldlearningframework AT claudiopersello buildingpolygonextractionfromaerialimagesanddigitalsurfacemodelswithaframefieldlearningframework
_version_	1718410529835319296

Building Polygon Extraction from Aerial Images and Digital Surface Models with a Frame Field Learning Framework

Ejemplares similares