Omnidirectional camera pose estimation and projective texture mapping for photorealistic 3D virtual reality experiences

Alessandro Luchetti*, Matteo Zanetti, Denis Kalkofen, Mariolino De Cecco

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

Modern applications in virtual reality require a high level of fruition of the environment as if it was real. In applications that have to deal with real scenarios, it is important to acquire both its three-dimensional (3D) structure and details to enable the users to achieve good immersive experiences. The purpose of this paper is to illustrate a method to obtain a mesh with high quality texture combining a raw 3D mesh model of the environment and 360° images. The main outcome is a mesh with a high level of photorealistic details. This enables both a good depth perception thanks to the mesh model and high visualization quality thanks to the 2D resolution of modern omnidirectional cameras. The fundamental step to reach this goal is the correct alignment between the 360° camera and the 3D mesh model. For this reason, we propose a method that embodies two steps: 1) find the 360° cameras pose within the current 3D environment; 2) project the high-quality 360° image on top of the mesh. After the method description, we outline its validation in two virtual reality scenarios, a mine and city environment, respectively, which allows us to compare the achieved results with the ground truth.

Original languageEnglish
Number of pages8
JournalActa IMEKO
Volume11
Issue number2
DOIs
Publication statusPublished - 2022

Keywords

  • camera pose estimation
  • enhanced comprehension
  • mesh reconstruction
  • Omnidirectional cameras
  • optimization

ASJC Scopus subject areas

  • Instrumentation
  • Mechanical Engineering
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Omnidirectional camera pose estimation and projective texture mapping for photorealistic 3D virtual reality experiences'. Together they form a unique fingerprint.

Cite this