Overcoming some of the issues in maintaining large urban area 3D models via a web browser

Abstract

This paper discusses recent work, in the VEPs Interreg project, that has been examining how to optimise operator based digital 3D modelling, of urban and rural environments, where the increasing availability of highly accurate LiDAR scanning offers appropriate remote sensed data. From this data bare earth digital terrain models (DTMs) can be processed semi-automatically, then draped with aerial imagery. Although both the resulting DTMs and geo-referenced aerial photography are of increasing detail and availability, buildings and similar structures are difficult to automatically extract at a level of detail that provides a credible sense of presence at street level. The LiDAR data itself is currently not captured often or extensively enough or at a sufficient level of detail that it could be relied on to keep VR analogues of real places up to date through an automated process alone, without additional modelling. Significant costs in modelling and updating urban settings to convey presence thus remain, be it through a combination of photogrammetry, CAD drawings and manual editing. It is argued that these costs are likely to limit large area 3D modelling to either broad brush overviews (as provided by Google Earth) or to credible presence models in those few locations where the high cost is justifiable, but in which, once the primary purpose of the model has been served, acceptance of the continuing cost of maintenance to reflect change is unlikely. Where updating of high presence models does happen this paper puts forward the view that currently, due to cost, it is likely to be piecemeal and take place over a long period of time. The paper describes one approach to reducing these costs by empowering heterogeneous groups of people to share the tasks of digitising and updating buildings that convey a credible sense of presence.