Deep learning for filtering the ground from ALS point clouds: A dataset, evaluations and issues

Qin, Nannan; Tan, Wei; Ma, Lingfei; Zhang, Dedong; Guan, Haiyan; Li, Jonathan

doi:10.1016/j.isprsjprs.2023.06.005

Cited by 12 publications

(9 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Case 2 is omitted here, but various aspects of this case are covered in Sec. 4. Actual induction of wavelengths λ from the recommended mean edge length l 0 to wavelengths remains an expert decision, solved case by case.…”

Section: Summary Of Resultsmentioning

confidence: 99%

“…Terrestrial and aerial laser scanning [1], satellite aperture array (SAR) tomography [2] and photogrammetry [3] each have different penetration, density and angular and depth accuracy. Also, large PC collections [4] are available for micro-topography feature capture, and there is material from different years and under different seasonal conditions. This results in a lot of quality tests and asks for systematical estimates of adequacy of PC sources, and in many cases, a degree of down-sampling before a possible fusion.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Highest informative density of a TIN model

Nevalainen,

Virjonen,

Salmivaara

et al. 2024

Preprint

View full text Add to dashboard Cite

A point cloud (PC) source is useful for micro-topography, if the shortest spatial wavelength it carries is enough for a particular registration task. This wavelengthvaries a lot due to different scanning methodologies and environmentalconditions. We propose an automated procedure for approximating the highest meaningful density of a PC in order to produce relatively noiseless ground models.This density limit can serve as a set target for existing thinningprocedures of large PC archives. Three methods to approximate the highest meaningful density are being compared:mean absolute error (MAE) of the triangulated irregular network (TIN) fit, minimum description length (MDL) and direct spatial wavelength registration (DWR). Each method is computationally expensive but can be subjected to a small set of local samples of a possibly large PC archive. Natural data samples were: a photogrammetric PC and a detail of it,and an aerial laser scan (ALS), all from the same forest site. The pipeline for finding a suggested minimum PC density has n log n complexity in terms of original PC size n. DWR found a wavelength range 5 ... 14 m from a synthetic case with the wavelength of 13 m. All methods give mean TIN edge length0.05 ... 0.1 m corresponding well with a published choice 0.2 m.

show abstract

Section: Summary Of Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Highest informative density of a TIN model

Nevalainen,

Virjonen,

Salmivaara

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…Designed explicitly to address the intrinsic complexities of high-resolution remote sensing imagery, PCCAU-Net is a deep learning model that features a unique multi-scale, multi-tiered strategy. Traditional deep learning approaches often encounter difficulties when confronted with diverse terrains, such as urban and rural landscapes [38]. In contrast, PCCAU-Net excels in road extraction tasks, executing them with both greater precision and robustness.…”

Section: Overview Of Pccau-netmentioning

confidence: 99%

“…multi-tiered strategy. Traditional deep learning approaches often encounter difficulties when confronted with diverse terrains, such as urban and rural landscapes [38]. In contrast, PCCAU-Net excels in road extraction tasks, executing them with both greater precision and robustness.…”

Section: Overview Of Pccau-netmentioning

confidence: 99%

PCCAU-Net: A Novel Road Extraction Method Based on Coord Convolution and a DCA Module

Xue,

Ren,

Yin

et al. 2024

Applied Sciences

View full text Add to dashboard Cite

In the domain of remote sensing research, the extraction of roads from high-resolution imagery remains a formidable challenge. In this paper, we introduce an advanced architecture called PCCAU-Net, which integrates Pyramid Pathway Input, CoordConv convolution, and Dual-Inut Cross Attention (DCA) modules for optimized performance. Initially, the Pyramid Pathway Input equips the model to identify features at multiple scales, markedly enhancing its ability to discriminate between roads and other background elements. Secondly, by adopting CoordConv convolutional layers, the model achieves heightened accuracy in road recognition and extraction against complex backdrops. Moreover, the DCA module serves dual purposes: it is employed at the encoder stage to efficiently consolidate feature maps across scales, thereby fortifying the model’s road detection capabilities while mitigating false positives. In the skip connection stages, the DCA module further refines the continuity and accuracy of the features. Extensive empirical evaluation substantiates that PCCAU-Net significantly outperforms existing state-of-the-art techniques on multiple benchmarks, including precision, recall, and Intersection-over-Union(IoU). Consequently, PCCAU-Net not only represents a considerable advancement in road extraction research, but also demonstrates vast potential for broader applications, such as urban planning and traffic analytics.

show abstract

“…Zhang et al [6] utilized a graph convolution network to filter point clouds in forest areas. Qin et al [34] assessed the performance of four 3D deep neural networks (PointNet++ [35], KPconv [36], RandLA-Net [37], and SCF-Net [38]) using the OpenGF dataset. The study results showed that learning-based pipelines outperformed classical filtering methods in most scenarios, particularly in forested environments with hybrid terrain.…”

Section: Introductionmentioning

confidence: 99%

A Multiscale Filtering Method for Airborne LiDAR Data Using Modified 3D Alpha Shape

Cao,

Wang,

et al. 2024

Remote Sensing

View full text Add to dashboard Cite

The complexity of terrain features poses a substantial challenge in the effective processing and application of airborne LiDAR data, particularly in regions characterized by steep slopes and diverse objects. In this paper, we propose a novel multiscale filtering method utilizing a modified 3D alpha shape algorithm to increase the ground point extraction accuracy in complex terrain. Our methodology comprises three pivotal stages: preprocessing for outlier removal and potential ground point extraction; the deployment of a modified 3D alpha shape to construct multiscale point cloud layers; and the use of a multiscale triangulated irregular network (TIN) densification process for precise ground point extraction. In each layer, the threshold is adaptively determined based on the corresponding α. Points closer to the TIN surface than the threshold are identified as ground points. The performance of the proposed method was validated using a classical benchmark dataset provided by the ISPRS and an ultra-large-scale ground filtering dataset called OpenGF. The experimental results demonstrate that this method is effective, with an average total error and a kappa coefficient on the ISPRS dataset of 3.27% and 88.97%, respectively. When tested in the large scenarios of the OpenGF dataset, the proposed method outperformed four classical filtering methods and achieved accuracy comparable to that of the best of learning-based methods.

show abstract

Deep learning for filtering the ground from ALS point clouds: A dataset, evaluations and issues

Cited by 12 publications

References 50 publications

Highest informative density of a TIN model

Highest informative density of a TIN model

PCCAU-Net: A Novel Road Extraction Method Based on Coord Convolution and a DCA Module

A Multiscale Filtering Method for Airborne LiDAR Data Using Modified 3D Alpha Shape

Contact Info

Product

Resources

About