Towards real-time photogrammetry

Abstract

The modern vision disciplines Computer Vision, Machine Vision, Robot Vision are in fact synonymous with Real-Time Photogrammetry. Promising market surveys, startling hardware developments and exciting algorithmic problems to be solved make this field a challenge for each photogrammetrist.

This overview paper is mainly concerned with a system's approach to real-time photogrammetry and a few hardware issues. Particular emphasis is put on the image acquisition problem, because this is the key to high-accuracy results. Some problems of current systems are highlighted and prospects for future directions indicated. A brief survey of present-day truly photogrammetric systems (systems with a substantial amount of photogrammetric expertise incorporated) is presented and their point-positioning accuracies, as obtained up to now, are reported.