A Rotation-Invariant Framework for Deep Point Cloud Analysis

Xianzhi Li, Ruihui Li, Guangyong Chen, Chi Wing Fu, Daniel Cohen-Or, Pheng Ann Heng

Research output: Contribution to journalArticlepeer-review

Abstract

Recently, many deep neural networks were designed to process 3D point clouds, but a common drawback is that rotation invariance is not ensured, leading to poor generalization to arbitrary orientations. In this paper, we introduce a new low-level purely rotation-invariant representation to replace common 3D Cartesian coordinates as the network inputs. Also, we present a network architecture to embed these representations into features, encoding local relations between points and their neighbors, and the global shape structure. To alleviate inevitable global information loss caused by the rotation-invariant representations, we further introduce a region relation convolution to encode local and non-local information. We evaluate our method on multiple point cloud analysis tasks, including (i) shape classification, (ii) part segmentation, and (iii) shape retrieval. Extensive experimental results show that our method achieves consistent, and also the best performance, on inputs at arbitrary orientations, compared with all the state-of-the-art methods.

Original languageEnglish
JournalIEEE Transactions on Visualization and Computer Graphics
DOIs
StateAccepted/In press - 2021

Keywords

  • Convolution
  • Feature extraction
  • Network architecture
  • Neural networks
  • Point cloud analysis
  • Shape
  • Task analysis
  • Three-dimensional displays
  • deep neural network
  • rotation-invariant representation

Fingerprint

Dive into the research topics of 'A Rotation-Invariant Framework for Deep Point Cloud Analysis'. Together they form a unique fingerprint.

Cite this