Connectivity-based convolutional neural network for classifying point clouds

作者：

Highlights：

• DenX-Conv is proposed to improve the accuracy of object classification.

• DenX-Conv also secures the connectivity of points from the raw point cloud.

• Stable feature learning was achieved by applying a DCN to a PointCNN's χ-Conv.

• DenX-Conv achieved a classification accuracy of 92.5% on the ModelNet40 dataset.

摘要

•DenX-Conv is proposed to improve the accuracy of object classification.•DenX-Conv also secures the connectivity of points from the raw point cloud.•Stable feature learning was achieved by applying a DCN to a PointCNN's χ-Conv.•DenX-Conv achieved a classification accuracy of 92.5% on the ModelNet40 dataset.

论文关键词：Convolutional neural networks,Delaunay triangulation,Dense connectivity,Neighbor connectivity,Point clouds classification

论文评审过程：Received 1 January 2020, Revised 1 September 2020, Accepted 16 October 2020, Available online 17 October 2020, Version of Record 30 January 2021.

论文官网地址：https://doi.org/10.1016/j.patcog.2020.107708