scholarly journals NormNet: Point-wise normal estimation network for three-dimensional point cloud data

2019 ◽  
Vol 16 (4) ◽  
pp. 172988141985753 ◽  
Author(s):  
Janghun Hyeon ◽  
Weonsuk Lee ◽  
Joo Hyung Kim ◽  
Nakju Doh
Keyword(s):  
Point Cloud ◽  
Nearest Neighbor ◽  
Three Dimensional ◽  
Local Feature ◽  
Data Sets ◽  
Point Cloud Data ◽  
K Nearest Neighbor ◽  
Cloud Data ◽  
Normal Estimation ◽  
Local Feature Extraction

In this article, a point-wise normal estimation network for three-dimensional point cloud data called NormNet is proposed. We propose the multiscale K-nearest neighbor convolution module for strengthened local feature extraction. With the multiscale K-nearest neighbor convolution module and PointNet-like architecture, we achieved a hybrid of three features: a global feature, a semantic feature from the segmentation network, and a local feature from the multiscale K-nearest neighbor convolution module. Those features, by mutually supporting each other, not only increase the normal estimation performance but also enable the estimation to be robust under severe noise perturbations or point deficiencies. The performance was validated in three different data sets: Synthetic CAD data (ModelNet), RGB-D sensor-based real 3D PCD (S3DIS), and LiDAR sensor-based real 3D PCD that we built and shared.

scholarly journals KVGCN: A KNN Searching and VLAD Combined Graph Convolutional Network for Point Cloud Segmentation

2021 ◽  
Vol 13 (5) ◽  
pp. 1003
Author(s):  
Nan Luo ◽  
Hongquan Yu ◽  
Zhenfeng Huo ◽  
Jinhui Liu ◽  
Quan Wang ◽  
...  
Keyword(s):  
Point Cloud ◽  
Nearest Neighbor ◽  
Semantic Segmentation ◽  
K Nearest Neighbor ◽  
Topological Graph ◽  
Convolutional Network ◽  
Cloud Data ◽  
Nearest Neighbor Searching ◽  
Point Cloud Segmentation ◽  
Local Feature Extraction

Semantic segmentation of the sensed point cloud data plays a significant role in scene understanding and reconstruction, robot navigation, etc. This work presents a Graph Convolutional Network integrating K-Nearest Neighbor searching (KNN) and Vector of Locally Aggregated Descriptors (VLAD). KNN searching is utilized to construct the topological graph of each point and its neighbors. Then, we perform convolution on the edges of constructed graph to extract representative local features by multiple Multilayer Perceptions (MLPs). Afterwards, a trainable VLAD layer, NetVLAD, is embedded in the feature encoder to aggregate the local and global contextual features. The designed feature encoder is repeated for multiple times, and the extracted features are concatenated in a jump-connection style to strengthen the distinctiveness of features and thereby improve the segmentation. Experimental results on two datasets show that the proposed work settles the shortcoming of insufficient local feature extraction and promotes the accuracy (mIoU 60.9% and oAcc 87.4% for S3DIS) of semantic segmentation comparing to existing models.

scholarly journals A GROSS ERROR ELIMINATION METHOD FOR POINT CLOUD DATA BASED ON KD-TREE

2018 ◽  
Vol XLII-3 ◽  
pp. 719-722 ◽  
Author(s):  
Q. Kang ◽  
G. Huang ◽  
S. Yang
Keyword(s):  
Point Cloud ◽  
Nearest Neighbor ◽  
Target Point ◽  
Point Cloud Data ◽  
K Nearest Neighbor ◽  
Cloud Data ◽  
Gross Error ◽  
Error Elimination ◽  
K Nearest Neighbor Algorithm ◽  
Key Steps

Point cloud data has been one type of widely used data sources in the field of remote sensing. Key steps of point cloud data’s pro-processing focus on gross error elimination and quality control. Owing to the volume feature of point could data, existed gross error elimination methods need spend massive memory both in space and time. This paper employed a new method which based on Kd-tree algorithm to construct, k-nearest neighbor algorithm to search, settled appropriate threshold to determine with result turns out a judgement that whether target point is or not an outlier. Experimental results show that, our proposed algorithm will help to delete gross error in point cloud data and facilitate to decrease memory consumption, improve efficiency.

scholarly journals PCA-Based Denoising Algorithm for Outdoor Lidar Point Cloud Data

Sensors ◽  
2021 ◽  
Vol 21 (11) ◽  
pp. 3703
Author(s):  
Dongyang Cheng ◽  
Dangjun Zhao ◽  
Junchao Zhang ◽  
Caisheng Wei ◽  
Di Tian
Keyword(s):  
Point Cloud ◽  
Nearest Neighbor ◽  
Principal Component ◽  
Noise Removal ◽  
Normal Vector ◽  
Point Cloud Data ◽  
Optimization Strategy ◽  
K Nearest Neighbor ◽  
Cloud Data ◽  
Run Speed

Due to the complexity of surrounding environments, lidar point cloud data (PCD) are often degraded by plane noise. In order to eliminate noise, this paper proposes a filtering scheme based on the grid principal component analysis (PCA) technique and the ground splicing method. The 3D PCD is first projected onto a desired 2D plane, within which the ground and wall data are well separated from the PCD via a prescribed index based on the statistics of points in all 2D mesh grids. Then, a KD-tree is constructed for the ground data, and rough segmentation in an unsupervised method is conducted to obtain the true ground data by using the normal vector as a distinctive feature. To improve the performance of noise removal, we propose an elaborate K nearest neighbor (KNN)-based segmentation method via an optimization strategy. Finally, the denoised data of the wall and ground are spliced for further 3D reconstruction. The experimental results show that the proposed method is efficient at noise removal and is superior to several traditional methods in terms of both denoising performance and run speed.

Three-Dimensional Reconstruction of Single Input Image Based on Point Cloud

2021 ◽  
Vol 87 (7) ◽  
pp. 479-484
Author(s):  
Yu Hou ◽  
Ruifeng Zhai ◽  
Xueyan Li ◽  
Junfeng Song ◽  
Xuehan Ma ◽  
...  
Keyword(s):  
Point Cloud ◽  
Large Scale ◽  
Three Dimensional ◽  
Data Sets ◽  
Point Cloud Data ◽  
Three Dimensional Reconstruction ◽  
Data Set ◽  
Cloud Data ◽  
Dimensional Reconstruction ◽  
Institute Of Technology

Three-dimensional reconstruction from a single image has excellent future prospects. The use of neural networks for three-dimensional reconstruction has achieved remarkable results. Most of the current point-cloud-based three-dimensional reconstruction networks are trained using nonreal data sets and do not have good generalizability. Based on the Karlsruhe Institute of Technology and Toyota Technological Institute at Chicago ()data set of large-scale scenes, this article proposes a method for processing real data sets. The data set produced in this work can better train our network model and realize point cloud reconstruction based on a single picture of the real world. Finally, the constructed point cloud data correspond well to the corresponding three-dimensional shapes, and to a certain extent, the disadvantage of the uneven distribution of the point cloud data obtained by light detection and ranging scanning is overcome using the proposed method.

scholarly journals HOLE-FILLING ALGORITHM FOR AIRBORNE LIDAR POINT CLOUD DATA

2020 ◽  
Vol XLII-3/W10 ◽  
pp. 739-745
Author(s):  
P. Liang ◽  
G. Q. Zhou ◽  
Y. L. Lu ◽  
X. Zhou ◽  
B. Song
Keyword(s):  
Boundary Point ◽  
Point Cloud ◽  
Nearest Neighbor ◽  
Nearest Neighbor Search ◽  
Airborne Lidar ◽  
Least Square ◽  
Point Cloud Data ◽  
K Nearest Neighbor ◽  
Cloud Data ◽  
Moving Least Square

Abstract. Due to the influence of the occlusion of objects or the complexity of the measured terrain in the scanning process of airborne lidar, the point cloud data inevitably appears holes after filtering and other processing. The incomplete data will inevitably have an impact on the quality of the reconstructed digital elevation model, so how to repair the incomplete point cloud data has become an urgent problem to be solved. To solve the problem of hole repair in point cloud data, a hole repair algorithm based on improved moving least square method is proposed in this paper by studying existing hole repair algorithms. Firstly, the algorithm extracts the boundary of the point cloud based on the triangular mesh model. Then we use k-nearest neighbor search to obtain the k-nearest neighbor points of the boundary point. Finally, according to the boundary point and its k-nearest neighbor point, the improved moving least squares method is used to fit the hole surface to realize the hole repair. Combined with C++ and MATLAB language, the feasibility of the algorithm is tested by specific application examples. The experimental results show that the algorithm can effectively repair the point cloud data holes, and the repairing precision is high. The filled hole area can be smoothly connected with the boundary.

scholarly journals Multi-Dimensional Underwater Point Cloud Detection Based on Deep Learning

Sensors ◽  
2021 ◽  
Vol 21 (3) ◽  
pp. 884
Author(s):  
Chia-Ming Tsai ◽  
Yi-Horng Lai ◽  
Yung-Da Sun ◽  
Yu-Jen Chung ◽  
Jau-Woei Perng
Keyword(s):  
Deep Learning ◽  
Point Cloud ◽  
Three Dimensional ◽  
Point Clouds ◽  
Training Data ◽  
Network Architectures ◽  
Point Cloud Data ◽  
Data Types ◽  
Raw Data ◽  
Cloud Data

Numerous sensors can obtain images or point cloud data on land, however, the rapid attenuation of electromagnetic signals and the lack of light in water have been observed to restrict sensing functions. This study expands the utilization of two- and three-dimensional detection technologies in underwater applications to detect abandoned tires. A three-dimensional acoustic sensor, the BV5000, is used in this study to collect underwater point cloud data. Some pre-processing steps are proposed to remove noise and the seabed from raw data. Point clouds are then processed to obtain two data types: a 2D image and a 3D point cloud. Deep learning methods with different dimensions are used to train the models. In the two-dimensional method, the point cloud is transferred into a bird’s eye view image. The Faster R-CNN and YOLOv3 network architectures are used to detect tires. Meanwhile, in the three-dimensional method, the point cloud associated with a tire is cut out from the raw data and is used as training data. The PointNet and PointConv network architectures are then used for tire classification. The results show that both approaches provide good accuracy.

ESTIMATING SURFACE NORMALS IN NOISY POINT CLOUD DATA

2004 ◽  
Vol 14 (04n05) ◽  
pp. 261-276 ◽  
Author(s):  
NILOY J. MITRA ◽  
AN NGUYEN ◽  
LEONIDAS GUIBAS
Keyword(s):  
Point Cloud ◽  
Smooth Surface ◽  
Smooth Curve ◽  
Local Information ◽  
Least Square ◽  
Neighborhood Size ◽  
Point Cloud Data ◽  
Cloud Data ◽  
Normal Estimation ◽  
Least Square Fitting

In this paper we describe and analyze a method based on local least square fitting for estimating the normals at all sample points of a point cloud data (PCD) set, in the presence of noise. We study the effects of neighborhood size, curvature, sampling density, and noise on the normal estimation when the PCD is sampled from a smooth curve in ℝ2or a smooth surface in ℝ3, and noise is added. The analysis allows us to find the optimal neighborhood size using other local information from the PCD. Experimental results are also provided.

Research on Generation Rules of Chest Width Point Curve Based on 3D Female Mannequin

2013 ◽  
Vol 796 ◽  
pp. 513-518
Author(s):  
Rong Jin ◽  
Bing Fei Gu ◽  
Guo Lian Liu
Keyword(s):  
Point Cloud ◽  
Three Dimensional ◽  
Upper Body ◽  
Secondary Development ◽  
Linear Regression Method ◽  
Point Cloud Data ◽  
Body Type ◽  
Cloud Data ◽  
Engineering Software ◽  
The Individual

In this paper 110 female undergraduates in Soochow University are measured by using 3D non-contact measurement system and manual measurement. 3D point cloud data of human body is taken as research objects by using anti-engineering software, and secondary development of point cloud data is done on the basis of optimizing point cloud data. In accordance with the definition of the human chest width points and other feature points, and in the operability of the three-dimensional point cloud data, the width, thickness, and length dimensions of the curve through the chest width point are measured. Classification of body type is done by choosing the ratio values as classification index which is the ratio between thickness and width of the curve. The generation rules of the chest curve are determined for each type by using linear regression method. Human arm model could be established by the computer automatically. Thereby the individual model of the female upper body mannequin modeling can be improved effectively.

Sign in / Sign up

Export Citation Format

Share Document