Expert system-based design of close-range photogrammetric networks

Designing an appropriate multi-station, convergent network is a prerequisite to the realization of high-accuracy photogrammetric measurement in industrial applications. Such networks are in practice designed by a simulation approach in which the expertise of the photogrammetrist is relied upon to overcome the complexity of the task. This article reports on investigations into the feasibility of an expert system solution to the automation of this design task. Major findings include recommendations for the representation of network design expertise, network topology and spatial information, the appropriate architecture for an expert system-based tool, development of conceptual models for camera placement and network diagnosis, and a computational model for camera placement. The prototype expert system CONSENS was built to test these concepts and models. Experiments with it confirm the feasibility of the approach. Prospects for application of this approach in photogrammetric practice, as well as limitations, are identified.