期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2004
卷号：XXXV Part B3
页码：140-144
出版社：Copernicus Publications
摘要：Traditional photogrammetry exploits stereoscopic image measurements to reconstruct the shape of objects. This technology is well- known and understood for over a century. Since the age of the stereo comparator human operators have been able to recognize the depth of a scene with remarkable accuracy. In principle, this has not changed during the time that image processing has been introduced into the photogrammetric production chain. Only a certain degree of automation has been derived, when matching and image correlation has replaced the human eyes. Nevertheless, we need to confess that shape from stereo is not very robust and that a time-consuming and therefore expensive control and verification process is needed after the automated reconstruction process. To overcome such problems, we argue that a process based on multiple image overlap is the appropriate solution. In the age of digital aerial cameras such multiple overlaps can be commanded without extra costs for additional images. A forward overlap of at least 80 % supports a robust and highly redundant measurement. Automation and verification starts to become an easy job, and based on the level of redundancy, we can expect a much more reliable means of automated production. In urban areas we have an additional benefit in the reduction or elimination of occlusions. In this paper we present several examples of multiple image matching, based on images taken with the new large format digital aerial camera UltraCam-D from Vexcel Imaging
