Open AccessA High-Resolution Satellite DEM Filtering Method Assisted with Building Segmentation
Author(s) -
Yihui Li,
Fang Gao,
Wentao Li,
Peng Zhang,
Yuan An,
Xing Zhong,
Yuwei Zhai,
Yongjian Yang
Publication year - 2021
Publication title -
photogrammetric engineering and remote sensing
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.483
H-Index - 127
eISSN - 2374-8079
pISSN - 0099-1112
DOI - 10.14358/pers.87.6.421
Subject(s) - orthophoto , digital elevation model , robustness (evolution) , satellite , computer science , segmentation , computer vision , terrain , remote sensing , artificial intelligence , filter (signal processing) , geography , engineering , cartography , biochemistry , chemistry , aerospace engineering , gene
Digital elevation model (DEM) filtering is critical in DEM production, and large-area meter-level resolution DEM is mainly generated from high-resolution satellite images. However, the current DEM filtering methods are mostly aimed at laser scanning data and tend to excessively remove ground points when processing a satellite digital surface model (DSM). To accurately filter out buildings and preserve terrain, we propose a DEM filtering algorithm using building segmentation results of orthophoto. Based on morphological filtering, our method estimates the probability of being a built-up area or mountains for DSM, and according to this probability the filtering parameters are adaptively adjusted. For robustness, our method performs the above filtering operation on DSM through a sliding-window approach, and finally the nonground points are determined by the votes of multiple filtering. Experiments against six representative data sets have shown that our method achieved superior perfor- mance than classical algorithms and commercial software.