Skip to main navigation Skip to search Skip to main content

Sight Guidance Enhanced VR Video Transmission

  • Beihang University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

This paper studies the enhancement of the tile-based multicasting transmission for virtual reality (VR) panoramic videos. A sight guidance scheme is proposed to aggregate the viewpoints of users, so as to increase the opportunity of multicasting and reduce the traffic load. Aimed at minimizing the weighted sum of average traffic load and users' watching preference derivation, we formulate a target guidance viewpoint optimization problem, and propose an approximate gradient descent algorithm to find a solution. Simulation results based on a real VR video dataset demonstrate the benefits brought by sight guidance, and show that compared with some heuristic sight guidance methods, the proposed algorithm has significant advantages in reducing traffic load and maintaining users' watching preference.

Original languageEnglish
Title of host publicationICSP 2022 - 2022 16th IEEE International Conference on Signal Processing, Proceedings
EditorsBaozong Yuan, Qiuqi Ruan, Shikui Wei, Gaoyun An
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages305-309
Number of pages5
ISBN (Electronic)9781665460569
DOIs
StatePublished - 2022
Event16th IEEE International Conference on Signal Processing, ICSP 2022 - Beijing, China
Duration: 21 Oct 2022 → 24 Oct 2022

Publication series

NameInternational Conference on Signal Processing Proceedings, ICSP
Volume2022-October
ISSN (Print)2164-5221
ISSN (Electronic)2164-523X

Conference

Conference16th IEEE International Conference on Signal Processing, ICSP 2022
Country/TerritoryChina
CityBeijing
Period21/10/22 → 24/10/22

Keywords

  • Virtual reality (VR)
  • panoramic video
  • sight guidance
  • traffic load
  • user preference

Fingerprint

Dive into the research topics of 'Sight Guidance Enhanced VR Video Transmission'. Together they form a unique fingerprint.

Cite this