Abstract
Geometric feature acts as an important role in point cloud shape classification tasks. Previous methods have proved that the geometric information of point clouds effectively improves the classification accuracy. Mo-Net firstly introduced geometric moments into point cloud shape classification, which, to fit the form of second order geometric moments, extends the number of input channels from three to nine. Unfortunately, similar to PointNet, Mo-Net cannot capture the local structures. To address this issue, we propose a graph geometric moments convolution neural network (GGM-Net), which learns local geometric features from geometric moments representation of a local point set. The core module of the GGM-Net is to learn features from geometric moments (termed as GGM convolution). Specifically, the GGM convolution learns point features and local features from the first and second order geometric moments of points and its local neighbors, respectively, and then combines these features by using an addition operation. In this way, a geometrical local representation about points is obtained, which leads to much surface geometry awareness and robustness. Equipped with the GGM convolution, GGM-Net, a simple end-to-end architecture, is developed to achieve a competitive accuracy on the benchmark dataset ModelNet40 and perform more efficiently in terms of memory and computational complexity.
Highlights
With the development of sensor technology, it is much easier to acquire 3D point cloud data than before, which promotes a wide range of the related applications, including automatic drive [1]–[3], robotics [4], urban point cloud labeling [41], [43], large scale scene understanding [42], [44], [45] and simultaneous localization and mapping (SLAM) [5]
We propose a graph geometric moments convolution neural network (GGM-Net) to learn local geometric features for classification accuracy improvement
POINT BASED METHODS In recent years, PointNet [30] pioneers the route of directly applying the standard convolution operation on unstructured raw 3D point clouds
Summary
With the development of sensor technology, it is much easier to acquire 3D point cloud data than before, which promotes a wide range of the related applications, including automatic drive [1]–[3], robotics [4], urban point cloud labeling [41], [43], large scale scene understanding [42], [44], [45] and simultaneous localization and mapping (SLAM) [5]. Mo-Net achieved a better 3D shape classification performance by using the higher order geometric moments representation of points as the input, it, similar to PointNet, only learns the features independently from each point, which neglects local structures. To address this issue, we propose a graph geometric moments convolution neural network (GGM-Net) to learn local geometric features for classification accuracy improvement. A novel learn-from-geometric-moments convolution operator, called GGM convolution, is proposed It can explicitly encode the local geometric structure of a point set, resulting in surface geometry awareness and robustness;. Extensive experiments demonstrate that GGM convolution has good transferability and superiority, and several hierarchical architectures equipped with GGM convolution achieve better performance
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
