Image-based modelling for augmenting reality

Abstract

The interaction between real and synthetic geometry is fundamental to Augmented Reality. Portraying such interactions in a manner that is convincing to the user requires 3D models of the shape of the real objects involved. The large-scale application of Augmented Reality technologies will thus require practical methods for generating 3D models of real objects. These methods will need to be fast, flexible, and capable of operating in-situ in order to generate models in unforeseen environments. Image-based modelling offers a means of creating such models by direct analysis of an image set. This paper describes two approaches to image-based modelling for Augmented Reality, and argues that technologies of this type are critical to the future of the domain.