Rich point clouds in virtual globes - A new paradigm in city modeling?

In this paper, we discuss the emergence of a new paradigm in 3D city modeling, which we will call the rich point cloud paradigm and which is largely based on dense and semantically rich 3D point clouds. We first present existing modeling paradigms, namely the explicit modeling of 3D geometry and image-based city modeling, and discuss their strengths and shortcomings. Key elements of the new modeling paradigm include the following: (1) treatment of 3D point clouds with increasingly rich geometric, radiometric and semantic properties as actual urban 3D models and not just as an input to the traditional 3D modeling process, (2) potential for fully automatically capturing and generating large scale 3D city models, (3) capturing rich detail of the micro- to meso-scale environment from a ground-based 'human' perspective supporting numerous drive-through and walk-through application scenarios and (4) possibility to start from fully automatically derived simple visual models and to gradually enhance and enrich them on an as-needed basis. We contend that virtual globes have a key role in fully exploiting the potential of the new paradigm. These arguments are supported with the implementation of a proof of concept prototype based on the i3D virtual globe technology and with initial observations from tests with real-world data sets. Finally, we identify a number of future research issues and challenges that are triggered by the new paradigm. (C) 2010 Elsevier Ltd. All rights reserved.