Spherical Image Processing for Immersive Visualisation and View Generation

Abstract

This research presents the study of processing panoramic spherical images for immersive visualisation of real environments and generation of in-between views based on two views acquired. For visualisation based on one spherical image, the surrounding environment is modelled by a unit sphere mapped with the spherical image and the user is then allowed to navigate within the modelled scene. For visualisation based on two spherical images, a view generation algorithm is developed for modelling an indoor manmade environment and new views can be generated at an arbitrary position with respect to the existing two. This allows the scene to be modelled using multiple spherical images and the user to move smoothly from one sphere mapped image to another one by going through in-between sphere mapped images generated.