On Training Deep-learning Models for Removing Airborne-Particle Points from SPAD LiDAR Multi-Echo Point Clouds

Tzu Hsien Sang, Yu Chen Lin, Yu Chan Hsiao

Research output: Contribution to journalArticlepeer-review

Abstract

SPAD (Single Photon Avalanche Diode) LiDAR is susceptible to be disrupted by adverse operating conditions, such as fog, haze or dust. Thereby the ability of LiDAR-based scene understanding would be significantly affected. In the published literatures, efforts to eliminate these effects are either based on processing histograms or processing point clouds. In this paper, we study the task of removing airborne-particle points from SPAD LiDAR multi-echo point clouds. To be more specific, the deep-learning approach is chosen, due to its promising performance and expandability. We focus on the issue of training deep-learning models for the task, especially when only datasets from limited scenarios and different types of LiDAR are available. A training strategy mainly based on data preparation and dataset augmentation, together with tweaks made on the neural network model, is shown to be effective in achieving the training task.

Original languageEnglish
Pages (from-to)1-4
Number of pages4
JournalIEEE Sensors Letters
DOIs
StateAccepted/In press - 2023

Keywords

  • airborne particle
  • Atmospheric modeling
  • deep learning
  • fog
  • Laser radar
  • LiDAR
  • Multi-echo point cloud
  • Point cloud compression
  • Sensors
  • Single-photon avalanche diodes
  • SPAD
  • Task analysis
  • Training

Fingerprint

Dive into the research topics of 'On Training Deep-learning Models for Removing Airborne-Particle Points from SPAD LiDAR Multi-Echo Point Clouds'. Together they form a unique fingerprint.

Cite this