الفهرس Only 14 pages are availabe for public view
 |  | from | 160 |  |  |
| | | from | 160 | | |
Abstract
Remote sensing data are widely used nowadays in map production, management of natural resources, and updating of natural and man-made changes which occur in land cover and land
use.
Geometric accuracy of space imagery plays a basic role in the accuracy of produced maps. Production of maps from space images undergoes many procedures and corrections on point positioning using Ground Control Points (GCPs).
In this thesis some mathematical methods to enable transformation of coordinates and rectification of images in order to obtain acceptable accuracy are introduced. The mathematical model that was adopted in this research aims at the determination of the best GCP-point scam regarding accuracy and distribution. These procedures were performed through a computer program which was based on the integration of three commercial sottwares; viz Matlab, Visual Basic and Excel. This program was used to study three images of three study-areas; namely;
• Alexandria image acquired by IRS-IC with geometric accuracy 5.8 m.
• Moneeb image acquired by IKONOS with geometric accuracy 4 m. • Zamalek image acquired by SPIN-2 with geometric accuracy 2 m.
For all study-areas two sources of GCPs were adopted; firstly GPS observations and secondly cadastral maps of scale 1:5000 which were acquired by air photo/;,’Taphy in 1994. In order to unify the coordinate system of the two sources it was necessary to carry out a transformation operation. A program was designed using a Visual Basic for this purpose which carries out the transformation between Helmert 1906 system (maps) and international system WGS84. In this program the seven transformation Molodensky parameters were determined.
Several researchers have introduced estimated values of Molodensky parameters. A large discrepancy• was observed between the previously determined values and present ones (several hundred meters). It was found that the reason for this was caused by a mistake in the grid system of the version of maps of scale 1:5000.
Rectification of space images to produce maps is usually based on using a mathematical transformation model; e.g. Affine or Polynomial model. In this thesis a procedure was suggested for this transformation which is based on the technolo!:,’Y of Artificial Neural Networks. This modern technology have been in use various disciplines for some times but never before in such an application.