From visual semantic parameterization to graphic visualization

Abstract

Visualizing natural language description is a difficult and complex task. When dealing with the process of generating images from natural language descriptions, we firstly should consider the real world and find out what key visual information can be extracted from the sentences which represents the most fundament concepts in both virtual and real environments. In this paper, we present the result of a prototype system called 3DSV (3D Story Visualiser) that generates a virtual scene by using simplified story-based descriptions. In particular, we describe the methodology used to parameterize the visual and describable words into XML formatted data structure. Then we discuss how to interpret the parameterized data and create an interactive real-time 3D virtual environment.