Photogrammetric reduction and analysis of real and simulated SPOT imageries
Abstract
The photogrammetric reduction of the SPOT satellite time-dependent imagery requires a special and sophisticated mathematical model that can fully exploit the many capabilities of SPOT imagery. A mathematical model based on collinearity equations with Eulerian orbital parameters has been developed where adjustment is made using unified least squares. Orbit parameters are used to describe the nominal satellite path, while positional and angular deviations from the nominal path are modeled as power series of time. The mathematics and techniques involved in collection and preprocessing of image coordinates, ground coordinates, and ancillary data are described. Four implementations of the mathematical model have been created to perform various photogrammetric reduction applications. The main implementation is block adjustment, in which simultaneous triangulation of a block of SPOT frames, taken from one or more orbits, is performed. The other three implementations deal with a single image frame at a time; these are: (1) Single ground point intersection to calculate the horizontal ground coordinates of a point from its image coordinates and ground elevation; (2) Single image point intersection to calculate image coordinates of a point using its ground coordinates; and (3) Single image frame resection to recover the adjusted parameters using image and ground coordinates of a sufficient number of known points. Using these implementations, simulated data were created and used to analyze the interaction among model variables (image coordinates, ground coordinates, and other model parameters) and the effect of their weights, and to study model performance at critical cases. Results are evaluated using root mean square error (RMS) on either the ground, or in image space where appropriate. Fourteen real SPOT stereo frames, one triplet, and two pairs of stereo strips are used to validate the mathematical model and provide conclusions on its performance and versatility. Several experiments were made with simulated data, but many more experiments were performed using real situations. With about 12 thousand experiments, mostly with real data, the mathematical model through its four implementations has proven to be a powerful tool for analysis of SPOT imageries. Sub-pixel accuracy using both simulated and real data has been achieved. Average RMS errors along Easting, Northing, and Elevation for the stereopairs used are (10.3, 9.8, 8.9) meters, respectively, using 6 control points. With only 3 control points, the results are (11.8, 10.7, 10.0) meters. Results of 4-frame strip adjustment with 6 control points are (15.6, 12.1, 6.2) meters, while with 3 control points they are (16.5, 12.3, 7.4) meters.
Degree
Ph.D.
Advisors
Mikhail, Purdue University.
Subject Area
Civil engineering|Earth
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.