An implementation method of multi-scale point cloud classification based on graph convolution
An implementation method and multi-scale technology, which can be used in instruments, biological neural network models, computing, etc., to solve problems such as lack of multi-scale features, and achieve the effect of improving accuracy and enriching reference features.
Active Publication Date: 2021-10-01
HUBEI UNIV OF TECH
View PDF4 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
However, this method can only capture edge features in a limited neighborhood and lacks multi-scale features
Therefore, it is still challenging to further improve the performance of 3D point cloud classification
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0056] refer to figure 1 , figure 2 and image 3 , which shows that the implementation method of multi-scale point cloud classification based on graph convolution of the present invention. The specific implementation steps include:
[0057] Step 1, improve the KNN proximity algorithm to obtain k points sampled at equal intervals in different scales, so as to construct M-KNN graphs of different scales of the point cloud set:
[0058] A D-dimensional point cloud with n points, expressed as:
[0059]
[0060] where X represents a set of point clouds, x i Represents each point, n represents the number of points in the point cloud collection, and D represents the dimension of the point cloud data.
[0061] Because, only the position information of the point cloud is used in the present invention, that is, D=3. Therefore, each point contains only its 3D coordinates, ie:
[0062] x i =(x i ,y i ,z i ) (2)
[0063] Calculate the Euclidean distance between points in the ...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention belongs to the technical field of three-dimensional point cloud classification, and discloses a multi-scale three-dimensional point cloud classification realization method based on graph convolution. Its purpose is to obtain more abundant neighborhood information by extracting the edge features of different scales of the point cloud collection, so as to improve the accuracy of point cloud classification. The specific steps are as follows: Step 1, through the improved KNN proximity algorithm, obtain k points sampled at equal intervals in different scales, so as to construct M-KNN graphs of different scales of the point cloud set; Step 2, use M-KNN graphs to build The EdgeConv (edge convolution layer) module realizes the extraction of point cloud edge features. Step 3, use the edge convolution layer to build a multi-scale point cloud classification network model, use the ModelNet40 dataset to train the network, and obtain the final point cloud classification network.
Description
technical field [0001] The invention belongs to the technical field of three-dimensional point cloud classification, and in particular relates to a multi-scale three-dimensional point cloud classification realization method based on graph convolution. Background technique [0002] 3D point cloud classification refers to the process of extracting artificial or natural geographic elements from complex unordered point clouds with information such as 3D spatial coordinates. In recent years, with the development of 3D laser scanning technology, the acquisition of 3D point cloud data has become fast and cheap. It has a wide range of applications in areas such as unmanned driving, robotics, and indoor scene detection and recognition. 3D point cloud classification has become a research hotspot in the field of computer vision. [0003] The traditional point cloud classification method is to manually design a series of features, and then use a suitable classifier to directly classify...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/62G06N3/04
CPCG06N3/045G06F18/214G06F18/24147
Inventor 张正文胡永东巩朋成张林让李婕潘懋舜柯凡余梦婕
Owner HUBEI UNIV OF TECH