Comprehensive Review of Deep Learning-Based 3D Point Cloud Completion Processing and Analysis

Fei, Ben; Yang, Weidong; Chen, Wen-Ming; Li, Zhijun; Li, Yikang; Ma, Tao; Hu, Xing; Ma, Ligeng

doi:10.48550/arxiv.2203.03311

Cited by 3 publications

(5 citation statements)

References 103 publications

(158 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The approximate calculation can be performed for the curvature according to the method in He et al [12], as shown in Eq. (7).…”

Section: Adaptive Neighborhood Constructionmentioning

confidence: 99%

“…For the 3D point cloud model, the feature line is the orderly connection of a series of feature points [37]. As there is no topological connection among the point cloud data itself, together with the problems such as uneven sampling, noise, and missing data, further discussion and research are still required on how to extract the feature points of the point cloud model quickly and with high quality [7,30].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Feature Extraction Algorithm Based on Optimal Neighborhood Size

Jia

Pang

Xie

2022

Circuits Syst Signal Process

View full text Add to dashboard Cite

The feature extraction of a point cloud fragment model is the basis of fragment splicing, which provides the technical support for research on the segmentation, splicing, and restoration of fragment surfaces. High-quality feature extraction, however, is a complicated process due to the diversity of the surface information of a fragment model. For this subject, a high-efficient point cloud feature extraction method was proposed to address a new method for extracting feature lines. First, the projection distance feature of the point cloud model was calculated to identify the potential feature points. Furthermore, the local information of the possible feature points was used to construct the adaptive neighborhoods for identifying the feature points based on neighborhoods of the model. The clustering fusion of the feature points was proposed according to the discrimination threshold values of the feature points. Finally, the Laplace operator was utilized to refine and connect the feature points to form smooth feature lines. The experimental results showed that the proposed method was automatic, highly efficient, and with good adaptability that could effectively extract the detailed features and construct the complete feature lines. Moreover, results showed that the provided framework could extract the features of simple structure models and be feasible to a certain extent for fragment models with abundant features.

show abstract

“…The approximate calculation can be performed for the curvature according to the method in He et al [12], as shown in Eq. (7).…”

Section: Adaptive Neighborhood Constructionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A Feature Extraction Algorithm Based on Optimal Neighborhood Size

Jia

Pang

Xie

2022

Circuits Syst Signal Process

View full text Add to dashboard Cite

show abstract

“…Wite the development of 3D sensors such as LiDAR (Light Detection and Ranging) sensors [1], [2] and depth cameras, point cloud, as the most common type of data representing 3D information of objects, gradually occupy an important position in computer vision [3], [4], autonomous driving [5], [6] and etc. The accompanying tasks about point cloud clas-sification, segmentation and 3D target detection have become hot research topics.…”

Section: Introductionmentioning

confidence: 99%

“…Qi et al [25] apply the multiscale concept to the processing of point cloud projections and add a multi-resolution filtering module to enhance the extraction of information from the model. The multi-view-based models are capable of extracting texture features of point cloud and have achieved leading accuracy results on many datasets [3], among which PointView-GCN [15] achieves SOTA on ModelNet40. Another method of structuring point cloud is voxel-based, which converts point clouds into fixed-size voxel squares and uses the number of points within the squares as features.…”

Section: Introductionmentioning

confidence: 99%

To Drop or to Select: Reduce the Negative Effects of Disturbance Features for Point Cloud Classification From an Interpretable Perspective

Zhang

et al. 2023

IEEE Access

View full text Add to dashboard Cite

The perturbation features limit the performance of point cloud classification models both for clean point cloud and point cloud obtained from real-world. In this paper, we propose two methods to enhance models by reducing the negative impact of the nuisance features from the perspective of interpretability, namely, dropping the nuisance points before inputting the point cloud into models and adaptively selecting the important features during the training process. The former is achieved by saliency analysis of models, the perturbation points are those with low contribution to models. For each sample, dropping the low contribution points based on the saliency scores is equivalent to filtering the perturbation features. We design a generic framework for generating saliency maps for various models and datasets, and obtain empirical values for the number of dropped points on each set of them. Then we apply the unsupervised dropping process to improve the robustness of models. The latter is achieved by adaptive downsampling, and we design a multi-stage learnable class-attention-based downsampling module to replace the commonly used Farthest Point Sampling (FPS). As the training progresses, the downsampling module tends to select the common features for each category, thus eliminating the nuisance features to improve the learning efficiency of the model. For dropping points (DP), we generate saliency maps for PointNet&++, DGCNN and PointMLP on ModelNet40 and ScanObjectNN, PointNet+DP reaches an overall accuracy (OA) of 92.5% and 72% on ModelNet40 and ScanObjectNN, surpassing the original model by 3.4% and 5.3%, the OA of PointNet++ with DP on ModelNet40 object classification can be raised from 91.8% to 93.7%. For adaptive feature selection (AFS), PointMLP-elite+AFS reaches an OA of 92.5% and a mean accuracy (mAcc) of 72% on ScanObjectNN, surpassing the original model by 0.8% and 1%. It reaches the level of PointMLP with 6.3% of the number of parameters. Considering the difficulty of the deployment of deep models, PointMLP-elite+AFS is the most cost-effective classification model known on ScanObjectNN. INDEX TERMS Dropping points, Adaptive feature selection, Point cloud denoising, Point cloud classification VOLUME 4, 2016

show abstract

“…It has been employed to acquire geometries from small objects [Zeisl et al 2013] to cityscale infrastructure [Lai et al 2011] in various applications such as SLAM and self-driving cars. Although it is possible to acquire full observations via panoramic scanning in some scenarios [Kurkela et al 2021], very often, it requires a completion step as only partial scans can be obtained in many cases [Fei et al 2022].…”

Section: Introductionmentioning

confidence: 99%

Shape Completion with Points in the Shadow

Zhang,

Zhao,

Wang

et al. 2022

Preprint

View full text Add to dashboard Cite

Comprehensive Review of Deep Learning-Based 3D Point Cloud Completion Processing and Analysis

Cited by 3 publications

References 103 publications

A Feature Extraction Algorithm Based on Optimal Neighborhood Size

A Feature Extraction Algorithm Based on Optimal Neighborhood Size

To Drop or to Select: Reduce the Negative Effects of Disturbance Features for Point Cloud Classification From an Interpretable Perspective

Shape Completion with Points in the Shadow

Contact Info

Product

Resources

About