Single Image Augmented Reality Using Planar Structures in Urban Environments

Abstract

In this paper, we present an effective method for integrating 3D augmented reality graphics into single images taken in urban environments. Building facades usually contain a large number of parallel lines aligned along several principal directions. We make use of the images of these 3D parallel lines and their corresponding vanishing points to recover a number of 3D planes from single 2D images of urban environments and then use them to represent the spatial layout of a scene. 3D objects are then aligned with respect to these recovered planes to achieve realistic augmented reality effects. In experiments we applied the proposed method to implement augmented reality in images from a benchmark image dataset of urban environments.