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：2005
卷号：XXXVI-3/W36
出版社：Copernicus Publications
摘要：Automatic registration of image sequences has been a subject of research for many years, both in the photogrammetric and computer vision communities. As part of the automation, linear orientation methods are used to obtain approximations for a subsequent bundle adjustment solution. Linear solutions can be at time "too general" particularly in a sense that they mostly employ uncalibrated cameras, a fact leading to severely unstable results in most photogrammetric problems such as the case for the direct linear transformation (DLT) in a nearly flat terrain. Furthermore, to the best of our knowledge, none of them handle more than two or three images simultaneously without imposing several theoretical constraints that cannot be guaranteed in practical imaging missions. In this paper a sub-optimal linear solution for the exterior orientation parameters of image sequences is developed. The proposed method is demonstrated on an aerial image strip. The paper shows that the method successfully generates reliable and accurate approximations both for the orientation parameters as well as for tie point coordinates. For an automatic extraction of the latter, the Scale Invariant Feature Transform (SIFT) algorithm is applied
