Building identification and virtual object manipulation in mobile outdoor augmented reality

Abstract

This thesis explores building identification and virtual object manipulation in a mobile outdoor augmented reality environment. Building identification refers to the recognition of the identity of a building. Existing studies require additional information such as building images in advance or have constraints on the types of possible input images. Therefore, we propose a system for building identification using a building boundary map and smartphone sensors. It measures the position and the orientation of the user using smartphone sensors and finds the building by reducing the area of candidate buildings in a step-by-step manner. During the validation of our system, it identified the buildings with an accuracy of up to 83.3%. Next, Virtual object manipulation is a technology to interact with virtual objects to move or rotate. However, in outdoor augmented reality environments, manipulating virtual objects is more difficult than indoors because there are challenges in interacting with the virtual objects. Therefore, we propose VOAR, a system for manipulating a virtual object based on reference objects. VOAR utilizes pole-like objects as references and uses a cylindrical coordinates interface based on them to represent the relative position of the virtual object. Additionally, it augments the virtual object in Google Street View images from various viewpoints around the reference object to provide the location of the virtual object. We implement a demo app and conduct a user study. The results show that users were able to manipulate with an error reduced to 19.2% of the existing interface, experiencing greater ease and higher performance.