| Summary: | ABSTRACT
In August 1996 the National Space Development Agency of Japan (NASDA) launched its satellite called the Advanced Earth Observation Satellite (ADEOS) carrying a very high spatial resolution sensor named A VNIR (Advanced Visible and Near-Infrared Radiometer). This sensor is capable of recording the earth surface with pointing angle ± 40 degrees perpendicular to the satellite flight direction. These characteristics are particularly attractive to develop a three-dimensional model from a pair of overlapped AVNIR images taken from two adjacent paths. However, in order to be used to generate a representative Digital Elevation Model (DEM) the geometric performance of the data imaged by this sensor need to be evaluated. This research assesses such performance.
A pair of overlapped panchromatic AVNIR satellite images were used as samples. Coordinate transformation algorithms including Affine, Projective, and Polynomial were developed to recta the images. Ground control points (GCPs) acquired from topographic maps were used as input into those three algorithms. In this case, the Root Mean Square Error (RMSE) of each algorithms applied in the left and right images indicate the image geometric performance. Furthermore, this research attempts to determine the B/H ratio of the images. A mathematical model representing collinearity conditions among objects on the ground and their corresponding points on the image was used.
|
|---|