Segmenting Salient Objects in 3D Point Clouds of Indoor Scenes Using Geodesic Distances

In order to deal with the problem that some existing semantic segmentation networks for 3D point clouds generally have poor performance on small objects, a Spatial Eight-Quadrant Kernel Convolution (SEQKC) algorithm is proposed to enhance the ability of the network for extracting fine-grained features from 3D point clouds. As a result, the semantic segmentation accuracy of small objects in indoor scenes can be improved. To be specific, in the spherical space of the point cloud neighborhoods, a kernel point with attached weights is constructed in each octant, the distances between the kernel point and the points in its neighborhood are calculated, and the distance and the kernel points’ weights are used together to weight the point cloud features in the neighborhood space. In this case, the relationship between points are modeled, so that the local fine-grained features of the point clouds can be extracted by the SEQKC. Based on the SEQKC, we design a downsampling module for point clouds, and embed it into classical semantic segmentation networks (PointNet++, PointSIFT and PointConv) for semantic segmentation. Experimental results on benchmark dataset ScanNet V2 show that SEQKC-based PointNet++, PointSIFT and PointConv outperform the original networks about 1.35–2.12% in terms of MIoU, and they effectively improve the semantic segmentation performance of the networks for small objects of indoor scenes, e.g., the segmentation accuracy of small object “picture” is improved from 0.70% of PointNet++ to 10.37% of SEQKC-PointNet++.

Download Full-text

A Flexible Inference Machine for Global Alignment of Wall Openings

Remote Sensing ◽

10.3390/rs12121968 ◽

2020 ◽

Vol 12 (12) ◽

pp. 1968

Author(s):

Jiaqiang Li ◽

Biao Xiong ◽

Rongjun Qin ◽

Armin Gruen

Keyword(s):

Reconstruction Algorithm ◽

Point Clouds ◽

Data Driven ◽

Global Alignment ◽

3D Point Clouds ◽

Novel Approach ◽

Essential Components ◽

Indoor Scenes ◽

Planar Wall ◽

Wall Openings

Openings such as windows and doors are essential components of architectural wall surfaces. It is still a challenge to reconstruct them robustly from unstructured 3D point clouds because of occlusions, noises and non-uniformly distributed points. Current research primarily focuses on meliorating the robustness of detection and pays little attention to the geometric correctness. To improve the reconstruction quality, assumptions on the opening layout are usually applied as rules to support the reconstruction algorithm. The commonly used assumptions, such as the strict grid and symmetry pattern, however, are not suitable in many cases. In this paper, we propose a novel approach, named an inference machine, to identify and use flexible rules in wall opening modelling. Our method first detects and models openings through a data-driven method and then refines the opening boundaries by global and flexible rules. The key is to identify the global flexible rules from the detected openings, composed by various combinations of alignments. As our method is oblivious of the type of architectural layout, it can be applied to both interior wall surfaces and exterior building facades. We demonstrate the flexibility of our approach in both outdoor and indoor scenes with a variety of opening layouts. The qualitative and quantitative evaluation results indicate the potential of the approach to be a general method in opening detection and modelling. However, this data-driven method suffers from the existence of occlusions and non-planar wall surfaces.

Download Full-text

INDOOR MESH CLASSIFICATION FOR BIM

ISPRS - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences ◽

10.5194/isprs-archives-xlii-4-535-2018 ◽

2018 ◽

Vol XLII-4 ◽

pp. 535-539 ◽

Cited By ~ 1

Author(s):

L. S. Runceanu ◽

N. Haala

Keyword(s):

Low Cost ◽

Semantic Segmentation ◽

Point Clouds ◽

Initial Step ◽

Automatic Generation ◽

Benchmark Dataset ◽

3D Point Clouds ◽

Indoor Scenes ◽

3D Indoor Scenes

<p><strong>Abstract.</strong> This work addresses the automatic reconstruction of objects useful for BIM, like walls, floors and ceilings, from meshed and textured mapped 3D point clouds of indoor scenes. For this reason, we focus on the semantic segmentation of 3D indoor meshes as the initial step for the automatic generation of BIM models. Our investigations are based on the benchmark dataset ScanNet, which aims at the interpretation of 3D indoor scenes. For this purpose it provides 3D meshed representations as collected from low cost range cameras. In our opinion such RGB-D data has a great potential for the automated reconstruction of BIM objects.</p>

Download Full-text

UNSUPERVISED SEGMENTATION OF INDOOR 3D POINT CLOUD: APPLICATION TO OBJECT-BASED CLASSIFICATION

ISPRS - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences ◽

10.5194/isprs-archives-xliv-4-w1-2020-111-2020 ◽

2020 ◽

Vol XLIV-4/W1-2020 ◽

pp. 111-118 ◽

Cited By ~ 1

Author(s):

F. Poux ◽

C. Mattes ◽

L. Kobbelt

Keyword(s):

Point Cloud ◽

Computing Time ◽

Region Growing ◽

Point Clouds ◽

Unsupervised Segmentation ◽

Cloud Data ◽

3D Point Clouds ◽

Object Based ◽

Indoor Scenes ◽

Region Growing Segmentation

Abstract. Point cloud data of indoor scenes is primarily composed of planar-dominant elements. Automatic shape segmentation is thus valuable to avoid labour intensive labelling. This paper provides a fully unsupervised region growing segmentation approach for efficient clustering of massive 3D point clouds. Our contribution targets a low-level grouping beneficial to object-based classification. We argue that the use of relevant segments for object-based classification has the potential to perform better in terms of recognition accuracy, computing time and lowers the manual labelling time needed. However, fully unsupervised approaches are rare due to a lack of proper generalisation of user-defined parameters. We propose a self-learning heuristic process to define optimal parameters, and we validate our method on a large and richly annotated dataset (S3DIS) yielding 88.1% average F1-score for object-based classification. It permits to automatically segment indoor point clouds with no prior knowledge at commercially viable performance and is the foundation for efficient indoor 3D modelling in cluttered point clouds.

Download Full-text

Semantic segmentation networks of 3D point clouds for RGB-D indoor scenes

Twelfth International Conference on Machine Vision (ICMV 2019) ◽

10.1117/12.2557603 ◽

2020 ◽

Author(s):

Ya Wang ◽

Andreas Zell

Keyword(s):

Semantic Segmentation ◽

Point Clouds ◽

3D Point Clouds ◽

Indoor Scenes

Download Full-text

Real-Time Plane Segmentation and Obstacle Detection of 3D Point Clouds for Indoor Scenes

Computer Vision – ECCV 2012. Workshops and Demonstrations - Lecture Notes in Computer Science ◽

10.1007/978-3-642-33868-7_3 ◽

2012 ◽

pp. 22-31 ◽

Cited By ~ 5

Author(s):

Zhe Wang ◽

Hong Liu ◽

Yueliang Qian ◽

Tao Xu

Keyword(s):

Real Time ◽

Point Clouds ◽

Obstacle Detection ◽

3D Point Clouds ◽

Indoor Scenes

Download Full-text

Multi-scale Information Extraction Residual Network for 3D Point Clouds Classification

2020 Chinese Automation Congress (CAC) ◽

10.1109/cac51589.2020.9327126 ◽

2020 ◽

Author(s):

Yumei Li ◽

Shan Meng ◽

Daoyuan Liang

Keyword(s):

Information Extraction ◽

Point Clouds ◽

Residual Network ◽

Multi Scale ◽

3D Point Clouds

Download Full-text

Symmetry Analysis of Oriental Polygonal Pagodas Using 3D Point Clouds for Cultural Heritage

Sensors ◽

10.3390/s21041228 ◽

2021 ◽

Vol 21 (4) ◽

pp. 1228

Author(s):

Ting On Chan ◽

Linyuan Xia ◽

Yimin Chen ◽

Wei Lang ◽

Tingting Chen ◽

...

Keyword(s):

Cultural Heritage ◽

Unmanned Aerial Vehicle ◽

Point Cloud ◽

Geometric Model ◽

Digital Camera ◽

Point Clouds ◽

Symmetry Analysis ◽

3D Point Clouds ◽

Transmission Towers ◽

Aerial Vehicle

Ancient pagodas are usually parts of hot tourist spots in many oriental countries due to their unique historical backgrounds. They are usually polygonal structures comprised by multiple floors, which are separated by eaves. In this paper, we propose a new method to investigate both the rotational and reflectional symmetry of such polygonal pagodas through developing novel geometric models to fit to the 3D point clouds obtained from photogrammetric reconstruction. The geometric model consists of multiple polygonal pyramid/prism models but has a common central axis. The method was verified by four datasets collected by an unmanned aerial vehicle (UAV) and a hand-held digital camera. The results indicate that the models fit accurately to the pagodas’ point clouds. The symmetry was realized by rotating and reflecting the pagodas’ point clouds after a complete leveling of the point cloud was achieved using the estimated central axes. The results show that there are RMSEs of 5.04 cm and 5.20 cm deviated from the perfect (theoretical) rotational and reflectional symmetries, respectively. This concludes that the examined pagodas are highly symmetric, both rotationally and reflectionally. The concept presented in the paper not only work for polygonal pagodas, but it can also be readily transformed and implemented for other applications for other pagoda-like objects such as transmission towers.

Download Full-text

Using Least-Squares Residuals to Assess the Stochasticity of Measurements—Example: Terrestrial Laser Scanner and Surface Modeling

Engineering Proceedings ◽

10.3390/engproc2021005059 ◽

2021 ◽

Vol 5 (1) ◽

pp. 59

Author(s):

Gaël Kermarrec ◽

Niklas Schild ◽

Jan Hartmann

Keyword(s):

Least Squares ◽

Laser Scanner ◽

Real Data ◽

Point Clouds ◽

Statistical Testing ◽

Normal Vector ◽

Engineering Structures ◽

3D Point Clouds ◽

Least Squares Adjustment ◽

Correlation Parameters

Terrestrial laser scanners (TLS) capture a large number of 3D points rapidly, with high precision and spatial resolution. These scanners are used for applications as diverse as modeling architectural or engineering structures, but also high-resolution mapping of terrain. The noise of the observations cannot be assumed to be strictly corresponding to white noise: besides being heteroscedastic, correlations between observations are likely to appear due to the high scanning rate. Unfortunately, if the variance can sometimes be modeled based on physical or empirical considerations, the latter are more often neglected. Trustworthy knowledge is, however, mandatory to avoid the overestimation of the precision of the point cloud and, potentially, the non-detection of deformation between scans recorded at different epochs using statistical testing strategies. The TLS point clouds can be approximated with parametric surfaces, such as planes, using the Gauss–Helmert model, or the newly introduced T-splines surfaces. In both cases, the goal is to minimize the squared distance between the observations and the approximated surfaces in order to estimate parameters, such as normal vector or control points. In this contribution, we will show how the residuals of the surface approximation can be used to derive the correlation structure of the noise of the observations. We will estimate the correlation parameters using the Whittle maximum likelihood and use comparable simulations and real data to validate our methodology. Using the least-squares adjustment as a “filter of the geometry” paves the way for the determination of a correlation model for many sensors recording 3D point clouds.

Download Full-text