期刊
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
卷 202, 期 -, 页码 246-261出版社
ELSEVIER
DOI: 10.1016/j.isprsjprs.2023.06.005
关键词
Ground filtering; Deep learning; Point cloud dataset; Comparative evaluation
The capability of ALS in collecting high-precision point clouds over large areas makes it valuable for geospatial applications. Automated ground filtering (GF) is a fundamental and challenging step, and recent advancements in deep learning (DL) techniques provide a new solution for this problem. However, there is a scarcity of public 3D geospatial datasets, especially for landform-scale GF tasks. Comprehensive advancements in DL-based GF pipelines are achieved by publishing a large-scale GF dataset and evaluating the advantages of DL techniques through experimental comparisons with traditional methods.
The capability of partially penetrating vegetation canopy and efficiently collecting high-precision point clouds over large areas makes airborne laser scanning (ALS) a valuable tool for various geospatial applications. However, automated ground filtering (GF), one fundamental and challenging step for most ALS applications, has remained a widely researched yet unsolved problem for decades. The recent breakthroughs in supervised deep learning (DL) techniques, which rely on sufficient and high-quality labeled datasets, provide a new solution to better solve this problem. Unfortunately, public 3D geospatial datasets are scarce, especially for those tailored for the landform-scale GF task. Moreover, whether advanced deep neural networks (DNNs) can be well-scaled to the problem of GF remains an open question. To comprehensively advance the development of effective DL-based GF pipelines, we first publish an ultra-large-scale GF dataset built upon open-access ALS point clouds of four different countries worldwide, which covers over 47 km' and nine different terrain scenes. Then, multiple attractive advantages of DL techniques in GF are evaluated through extensive experimental comparisons with traditional GF methods on the presented dataset. Furthermore, we reveal several issues faced by generalizing existing advanced 3D DNNs into GF tasks with a series of in-depth experimental analyses. Finally, some promising directions for future research are suggested in response to the identified challenges. Our dataset, named OpenGF, is available at https://github.com/Nathan-UW/OpenGF.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据