ATTENTION: This is a web archive! The IMS Group was split up in 2018 and does not exist anymore. Recent work of former members can be found at the VR/AR Group and the Computer Vision Group.

Interactive Media Systems, TU Wien

Real-Time Visualization Pipeline for Dynamic Point Cloud Data

Thesis by Hansjörg Hofer

Supervision by Margrit Gelautz

Abstract

Current developments in sensor and computer vision technologies are pushing the boundaries of extended reality (XR) applications. Digital customizable avatars, mimicking human emotions and gestures, are soon to be replaced by true 3D capturings of humans in mixed reality environments. Recording action filled scenes to recreate and place them in virtual or remote environments promises more lifelike and immerse experiences, extending the possibilities of virtual reality (VR), augmented reality (AR) and mixed reality (MR) applications. Photogrammetry and depth ranging sensors are already capable of bringing rigid real-world objects to the virtual world, with astoundingly realistic results. In addition, with modern depth sensors entering the smartphone industry, the next steps for mobile 3D capturing and online reconstruction have already been taken. However, dynamic content with rapid changes in structure and topology is challenging to reconstruct, particularly for usage in real-time. This thesis presents a novel visualization pipeline, which facilitates the real-time reconstruction of dynamic point clouds in a widely used, flexible and powerful framework. We contribute a novel modular processing and rendering pipeline for the Unity3D game engine, a popular multi-platform engine for XR applications and mobile games. The implementation is capable of reconstructing photo-realistic 3D objects from dynamic depth sensor streams. Furthermore, the modular architecture allows for scalability and easy expandability. The quality of this novel visualization pipeline is determined through optical comparisons with classical visualization techniques, which show that those techniques are not easily distinguishable from the results of our work. Moreover, the exact execution time of the algorithms is measured for different point clouds to proof their real-time ability.

Reference

H. Hofer: "Real-Time Visualization Pipeline for Dynamic Point Cloud Data"; Supervisor: M. Gelautz; Institut für Softwaretechnik und interaktive Systeme, 2018; final examination: 10-03-2018.

BibTeX

Click into the text area and press Ctrl+A/Ctrl+C or ⌘+A/⌘+C to copy the BibTeX into your clipboard… or download the BibTeX.