A 3D Point Cloud Model Classification Method Based on Convolutional Neural Network

Abstract

The invention discloses a three-dimensional point cloud model classification method based on a convolutional neural network, comprising the steps of: S1, selecting Princeton ModelNet, aiming at ModelNet10 and ModelNet40 respectively, selecting a required number of models from the official website as training data and test data, and generating Training set and data set; S2, feature analysis on the point cloud model and construction of a classification framework; S3, ordering the point cloud; S4, two-dimensional image of the ordered point cloud data; S5, constructing a two-dimensional point-oriented CNN network for cloud images. For the first time, the present invention directly applies CNN in the image field to the classification of 3D point cloud models, and achieved classification accuracy rates of 93.97% and 89.75% respectively on ModelNet10 and ModelNet40, which are comparable to the current best methods. The experimental results fully show that, It is feasible to apply CNN in the image field to the classification of 3D point cloud models. The PCI2CNN proposed in this paper can effectively capture the 3D feature information of the point cloud model and is suitable for the classification of 3D point cloud models.

Description

technical field [0001] The invention relates to the technical fields of computer graphics, computer vision and intelligent recognition, in particular to a three-dimensional point cloud model classification method based on a convolutional neural network. Background technique [0002] With the rapid development of modern computer vision research, breakthroughs have been made in areas such as unmanned vehicles, autonomous robots, real-time SLAM technology and virtual 3D models, which has promoted the development of the usability of 3D point cloud data, and has also given birth to the Research on various applications of 3D point cloud data. Among them, the classification of point cloud data is the basis and key of various applied research. [0003] At present, deep learning technology has made breakthroughs in the field of image and speech recognition, which also provides a useful research direction for the classification of 3D models. However, the input that the deep learning...

AI Technical Summary

Problems solved by technology

However, these works all need to design a special network for the point cloud model, and it is impossible to apply the Convolution Neural Network (CNN), which has achieved great success in the field of image recognition, to the classification of point cloud models.

Images