Obtaining the optimal reduction of forecast models for space images

The article discusses the results of the study of classical models for the processing of aerospace images and the application of the regression modeling approach to solving problems of coordinate transformation using small samples of reference points on personnel and scanner aerospace images. Estimation of the parameters of mathematical models used to convert images on aerospace images into given cartographic projections is an important step, both in constructing the corresponding maps and in solving scientific and technical problems. Obtaining maps for various purposes for the surface of the Earth is impossible without solving the problem of the coordinate reference, which involves determining the coordinates of objects in a certain reference system. The coordinate reference problem, which is solved by aerospace images, does not fundamentally differ from astrophotographic one. Nevertheless, in some cases, especially when creating maps, the range of its application is much wider - not only geodetic points, but also many points of contours of various images (coastline, forests, etc.) can be transferred to the core network system.. A common and important procedure for processing measurements on aerospace images is to estimate the parameters of mathematical models used in particular for image conversion. Instead of the classical approach to parameter estimation, which provides for a rigidly fixed model, it is proposed to use the methodology of regression modeling implemented in the package “System of search for optimal regressions”. The effectiveness of this approach for solving the problem of coordinate transformation on aerospace images of the Earth's surface, new in its application, has been experimentally confirmed. Improving the accuracy of using this approach is provided by the procedure of structural identification, which involves the formation of a set of competing structures based on the original perspective model and the search for the optimal structure according to a given quality criterion.