Movsesian I. Computational methods of the selection of reference stars on the digital images

The object of research is a set of measurements of stars formed on a series of digital frames. The purpose of research - increase performance accuracy of the estimate of the angular positions of objects in space image data, due to the design and analysis of computer digital image processing techniques. Methods of research - the main results of the work are obtained on the basis of the theory of mathematical statistics, the statistical theory of testing multi-alternative hypotheses, and the estimation of parameters that made it possible to develop computational methods for private and complete preliminary identification; computational method of automatic selection of reference stars; a computational method for studying the accuracy of estimates of the angular positions of reference stars in space from image data. Theoretical and practical results - the developed computational methods of automatic selection of reference stars allow to significantly increase the accuracy of the estimation of the angular positions of objects, which is confirmed by the results of processing of CCD frames. Scientific novelty of the results: the first proposed a computational method of automatic selection of reference stars on digital images, which uses an expanded set of their selective features and a multi-stage increase in the order of the model of reduction of permanent plates with a simultaneous increase in the number of pairs used for their calculation, and allows the determination of the angular position objects in the space on the image data. The computational method of identifying measurements of objects in a series of digital images has been improved in the part of determining the initial approximation of the parameters of the pairwise correspondence between two sets of object measurements using the procedure of private preliminary identification, which makes it possible to increase the reliability of identifying measurements of objects on a series of digital images. The computational method of identifying the measurements of frames with the objects of the catalog has been improved in the part of determining the initial approximation under the conditions of essential uncertainty of their identification parameters with the help of the procedure of full preliminary identification, which makes it possible to increase the reliability of identifying the measurements of frames with catalog objects under conditions of essential uncertainty of their parameters. A method for studying the accuracy of measurements of the angular position of reference stars in space from image data has been developed further, in part of analyzing an expanded set of accuracy measures for the angular positions of reference stars from various subsamples of data. Dissertational work was carried out in accordance with the plan of research works of the department of electronic computers of the Kharkov National University of Radio Electronics within the framework of the number 307 "Dynamic intellectual analysis of sequences of fuzzy information under conditions of significant uncertainty based on hybrid systems of computational intelligence" (No. SR 0116U002539). The developed computational methods for processing digital images were introduced into the educational process at the Kharkiv National University of Radio Electronics, Odesa Astronomical Observatory (Odesa, Ukraine, MPC code - 583) and in the Space Research Laboratory of "Uzhgorod National University". The developed computational methods can be used in various software complexes for processing digital images for highly accurate determination of the angular positions of objects, for example, in programs for automated detection of small bodies of the solar system on a series of digital frames.