A point cloud classification and semantic segmentation method
A semantic segmentation and point cloud technology, which is applied in image analysis, image enhancement, instruments, etc., can solve problems such as difficult point cloud, sensor noise, rigid rotation of objects, etc., and achieve the effect of improving accuracy
Active Publication Date: 2022-07-12
GUANGDONG UNIV OF TECH
View PDF10 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
However, in the problem of point cloud data processing, there are problems such as object scanning being occluded, sensors containing noise, and objects rigidly rotating, making point clouds difficult to be effectively expressed by the above methods
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
[0031] The present invention proposes a method for object classification and semantic segmentation of 3D point cloud, which is based on the original 3D point cloud position information collected by 3D scanning equipment without special preprocessing such as voxelization or gridding. figure 1 Shown: extract features according to the point cloud data collected by the 3D scanning equipment, and discriminate the feature expression of the extracted feature points. If the feature expression confidence is high, it will be classified into the corresponding category; If it is lower, the point position information of the point and the adjacent points is introduced to re-establish the local similarity expression. Using graph theory to construct a network graph to classify local similarity, so as to improve the classification effect of point cloud; the classification results of the same category of point clouds determined by feature expression and local similarity expression are aggregated...
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 discloses a point cloud classification and semantic segmentation method, comprising: extracting features from collected point cloud data, discriminating and classifying the feature expressions of the extracted feature points, and classifying the feature points into high confidence and low confidence For feature points with high confidence, the category corresponding to the maximum value in the feature vector of feature points is taken as the category of feature points; for feature points with low confidence, the position information between feature points is used to establish the relationship between feature points. The similarity expression matrix of The feature points of the categories are aggregated and merged to achieve semantic segmentation of point clouds. The invention constructs a global feature similarity matrix by using the feature vector representing the feature information of a single point, and uses the Euclidean distance between the points to increase the association constraint between the points, thereby improving the accuracy of point cloud classification.
Description
technical field [0001] The invention relates to the field of three-dimensional environment segmentation, in particular to a point cloud classification and semantic segmentation method. Background technique [0002] Effective cognition and recognition of the 3D environment is the premise and foundation for robots to complete autonomous behavior, and point cloud classification and semantic segmentation are key steps in this field. Therefore, accurate classification and accurate segmentation of point clouds are crucial. At present, the common 3D object classification scheme usually completes object segmentation by first converting voxelization, surface meshing, or converting 3D point cloud into a form that is easy to express from multiple perspectives. [0003] With the update and upgrade of sensors and the development of big data, it is more and more convenient to obtain the point cloud of the space environment. However, in the problem of point cloud data processing, there a...
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): G06T7/10G06T7/136G06V10/26G06V10/764
CPCG06T7/10G06T7/136G06T2207/10028G06V10/267G06F18/24
Inventor 朱蕾陈炜楠何力管贻生张宏
Owner GUANGDONG UNIV OF TECH