Author:
Moradi Behrooz,Zoej Mohammad Javad Valadan,Yaghoobi Sayad,Yavari Somayeh
Abstract
Recently, linear features in remotely sensed imagery have gained much attention because of their unique characteristics compared to other control features. For georeferencing high-resolution satellite images, the observations in the mathematical equations (slope and y-intercept) of
the corresponding control lines in the two spaces are considered the same based on recent studies. However, the use of such assumptions causes error and reduces the accuracy of registration. The aim of this article is to present a methodology based on a quasi-observation assumption in the
mathematical equations in the process of georeferencing. Experimental results for IKONOS and GeoEye images over two different cities of Iran indicate that the quasi-observation assumption can increase the average registration accuracy up to 48.96% and 24.77% using 3D-affine and rational function
models, respectively. This improvement in accuracy increases the processing time by 31.48% over traditional approaches; however, the proposed methodology can be regarded as an efficient solution.
Publisher
American Society for Photogrammetry and Remote Sensing
Subject
Computers in Earth Sciences