Enhanced life-size holographic telepresence framework with real-time three-dimensional reconstruction for dynamic scene

• Main Author: Fadzli, Fazliaty Edora
• Format: Thesis
• Language: English
• Published: 2022
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://eprints.utm.my/id/eprint/101530/1/FazliatyEdoraPSC2022.pdf; http://eprints.utm.my/id/eprint/101530/; http://dms.library.utm.my:8080/vital/access/manager/Repository/vital:150557
• Institution: Universiti Teknologi Malaysia

Three-dimensional (3D) reconstruction has the ability to capture and reproduce 3D representation of a real object or scene. 3D telepresence allows the user to feel the presence of remote user that was remotely transferred in a digital representation. Holographic display is one of alternatives to discard wearable hardware restriction, it utilizes light diffraction to display 3D images to the viewers. However, to capture a real-time life-size or a full-body human is still challenging since it involves a dynamic scene. The remaining issue arises when dynamic object to be reconstructed is always moving and changes shapes and required multiple capturing views. The life-size data captured were multiplied exponentially when working with more depth cameras, it can cause the high computation time especially involving dynamic scene. To transfer high volume 3D images over network in real-time can also cause lag and latency issue. Hence, the aim of this research is to enhance life-size holographic telepresence framework with real-time 3D reconstruction for dynamic scene. There are three stages have been carried out, in the first stage the real-time 3D reconstruction with the Marching Square algorithm is combined during data acquisition of dynamic scenes captured by life-size setup of multiple Red Green Blue-Depth (RGB-D) cameras. Second stage is to transmit the data that was acquired from multiple RGB-D cameras in real-time and perform double compression for the life-size holographic telepresence. The third stage is to evaluate the life-size holographic telepresence framework that has been integrated with the real-time 3D reconstruction of dynamic scenes. The findings show that by enhancing life-size holographic telepresence framework with real-time 3D reconstruction, it has reduced the computation time and improved the 3D representation of remote user in dynamic scene. By running the double compression for the life-size holographic telepresence, 3D representations in life-size is smooth. It has proven can minimize the delay or latency during acquired frames synchronization in remote communications.