Comparison of 3D wire-frame models derived from various spatial data sources
Many techniques exist for 3D wire-frame model (WFM) extraction. These are well documented and many are successful for a given problem. The Defence Science and Technology Organisation’s Weapons Systems Division is interested in understanding 3D WFMs for applications in target acquisition by networked weapon systems, and in particular the mission planning aspects in the targeting cycle. The aim of this study is to assess, from the point of view of a weapon system’s performance, the similarities and differences between WFMs extracted from various spatial datasets. These datasets include high resolution aerial photography, high resolution digital aerial imaging, small format aerial photography and airborne laser scanning. Although the specific techniques used for the WFM extraction are not the focus of this paper, an overview of the processing required to extract each of the datasets is provided. The 3D WFMs have been compared with a reference dataset derived from ground-based, close range photogrammetry. Close range photogrammetry was chosen as the control as it has the highest spatial resolution, which correlates with a higher accuracy 3D WFM.
As anticipated, results indicate that the accuracy of the 3D WFM extraction process correlates to the spatial and spectral resolution of the data source. Airborne laser scanning is a case in point, where a 1m spatial resolution does not necessarily correlate to a high degree of accuracy in the extraction process, primarily because determining the exact location of the edges of buildings is difficult to achieve. This is well documented; however, future research in this area aims to determine what impact these errors are expected to have on weapon system effectiveness. Analysing the differences between the datasets and providing some underlying reasoning will allow for a more quantitative assessment of the impact of using these data for weapons targeting.