Object identification by a deictic representation for Cyber Annotation

Abstract

When we record the contents of conversations in a working space, there seems to be a problem that the other people are not able to identify the object which is represented by a deictic representation. In this paper, we propose a method to identify such objects. The proposed method estimates the object that is referenced by a pronoun in conversations, by using the distance information which pronouns have, and the position and orientation data of the speaker. We also developed an user interface using three-dimensional measurement of six markers, for the measurement of the speaker's head position and orientaion. Then we developed system for the proposed method.