Learning-Based Hole Detection in 3D Point Cloud Towards Hole Filling

Tabib, Ramesh Ashok; Jadhav, Yashaswini; Gani, Kiran; Patil, Ujwala; Mudenagudi, Uma

doi:10.1016/j.procs.2020.04.050

Cited by 14 publications

(4 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Most of the existing methods require manual setting of parameter thresholds after analyzing the property of the point cloud. In order to overcome this limitation, Tabib et al [24] proposed a deep learning framework to detect holes in the point cloud by directly taking the disordered point cloud as the input.…”

Section: Hole Detection Methods Based On Scattered Point Cloud Datamentioning

confidence: 99%

Object defect detection based on data fusion of a 3D point cloud and 2D image

Zhang¹,

Zhou²,

Liu³

et al. 2022

Meas. Sci. Technol.

View full text Add to dashboard Cite

In the process of defect detection, there could be interference factors such as poor image quality and missing point cloud data as a result of the complexity of the acquisition environment. Certain limitations can be found relying solely on point cloud data processing or image feature detection. Therefore, this paper tries a more intuitive and effective exploration. Firstly, an algorithm named hole boundary points detection of point cloud based on multi-scale principal component analysis is proposed, which can achieve the preliminary detection of hole boundary points while calculating the normal vector of each point in the point cloud. Then the boundary contour of each hole is constructed by a polygon growth algorithm. Finally, we use the complementary information of three-dimensional (3D) point cloud and two-dimensional (2D) image to explore the origin of holes and realize the "true" and "false" classification of holes. The experimental results show that our algorithm can successfully detect point cloud holes and can also distinguish them from object defects, providing data support for subsequent holes filling and defect measurement.

show abstract

Section: Hole Detection Methods Based On Scattered Point Cloud Datamentioning

confidence: 99%

Object defect detection based on data fusion of a 3D point cloud and 2D image

Zhang¹,

Zhou²,

Liu³

et al. 2022

Meas. Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…Based on the high similarity between the hole and its neighborhood data, the existing methods perform interpolation repair on the hole by analyzing the data's spatial correlations and using them to fit the surface model where the hole is located [7]. The commonly used surface fitting algorithms include the moving least squares (MLS) method [8], B-spline curve method [9][10][11], radial basis function interpolation (RBF) [5], and deep learning method [12].…”

Section: Related Workmentioning

confidence: 99%

Hole Filling of Single Building Point Cloud Considering Local Similarity among Floors

Min

Wang

et al. 2022

Remote Sensing

View full text Add to dashboard Cite

In the process of acquiring a point cloud, due to the shielding of the building itself and other ground objects (such as vegetation), the collected point cloud cannot cover the building facades surface uniformly and completely, and can produce several holes of different sizes. The existing hole-filling methods cannot obtain satisfying repaired results. To solve this problem, this paper proposes a new hole-filling approach considering the local similarity among floors. The proposed approach first detects holes in the building facade surface and then identifies the building floors; next, according to the fact that building data are similar in the same position on different floors, the holes of the building facade point cloud can be filled using the data of different floors. The study examines three large buildings to verify the proposed method. The experimental results show that the proposed method outperforms the state-of-the-art filling methods; the filling results of the proposed method are closer to the real shape and can be smoothly connected with the original building point cloud. Moreover, the proposed method has greater accuracy when comparing the root mean square error (RMSE).

show abstract

“…Unfortunately, the threshold selections of these estimating indicators depend highly on the local distribution characteristics of scattered point clouds, such as their uniformities and densities. Thus, it is totally non-intuitive and very difficult to select an optimal threshold that can guarantee a fine extraction accuracy for unorganized point clouds with highly uneven and strongly random distribution [ 11 ]. Although multiple different estimating indicators can be commonly adopted to conduct the boundary extractions on various complex-shaped point clouds for higher accuracies, their threshold selections are still very complex and challenging in practices.…”

Section: Introductionmentioning

confidence: 99%

A Novel Type of Boundary Extraction Method and Its Statistical Improvement for Unorganized Point Clouds Based on Concurrent Delaunay Triangular Meshes

Wang

Feng

et al. 2023

Sensors

View full text Add to dashboard Cite

Currently, three-dimensional (3D) laser-scanned point clouds have been broadly applied in many important fields, such as non-contact measurements and reverse engineering. However, it is a huge challenge to efficiently and precisely extract the boundary features of unorganized point cloud data with strong randomness and distinct uncertainty. Therefore, a novel type of boundary extraction method will be developed based on concurrent Delaunay triangular meshes (CDTMs), which adds the vertex-angles of all CDTMs around a common data point together as an evaluation index to judge whether this targeted point will appear at boundary regions. Based on the statistical analyses on the CDTM numbers of every data point, another new type of CDTM-based boundary extraction method will be further improved by filtering out most of potential non-edge points in advance. Then these two CDTM-based methods and popular α-shape method will be employed in conducting boundary extractions on several point cloud datasets for comparatively analyzing and discussing their extraction accuracies and time consumptions in detail. Finally, all obtained results can strongly demonstrate that both these two CDTM-based methods present superior accuracies and strong robustness in extracting the boundary features of various unorganized point clouds, but the statistically improved version can greatly reduce time consumption.

show abstract

Learning-Based Hole Detection in 3D Point Cloud Towards Hole Filling

Cited by 14 publications

References 13 publications

Object defect detection based on data fusion of a 3D point cloud and 2D image

Object defect detection based on data fusion of a 3D point cloud and 2D image

Hole Filling of Single Building Point Cloud Considering Local Similarity among Floors

A Novel Type of Boundary Extraction Method and Its Statistical Improvement for Unorganized Point Clouds Based on Concurrent Delaunay Triangular Meshes

Contact Info

Product

Resources

About