Kozhukhov O. Computational methods of image processing for faint objects' position estimation

Object of research: data processing for objects' detection on digital frames. Research objective - improvement of accuracy in faint objects' detection on digital frames using effective computational methods of digital images processing. Research methods - Main results were obtained using: statistic theory checking hypothesis and parameters' evaluation that helped to synthesize method of objects' coordinates determination on a digital frame; numerical methods and mathematical statistic methods that permitted to develop a method of preliminary selection of signals received from faint objects with nonzero parameters of apparent movement; method of statistic modeling and full-scale data processing proving credibility of obtained theoretical results. Equipment - the personal computer. Theoretical and practical results - developed computational methods can be used in different software for digital images processing, preliminary selection and location evaluation of faint objects (for example within software for small Solar System bodies' (asteroids and comets) automatic detection on series of digital frames). Science novelty: updating of iterative method of objects' location evaluation on a digital image including obstacles with unknown parameters that permits to reduce mean square deviation of faint object coordinates' estimation comparing with traditional methods based upon object's image approximation by Gaussian model; improvement of method of frames' stacking accumulating signals received from an faint object with nonzero apparent movement that permits to increase probability for objects detection using "areal" approach and matrix of blurring with optimal values of elements; for the first time we represented a method of preliminary selection of objects' on digital images that permits to decrease computational costs for objects' coordinates estimation on digital images with no loss in quality of detection comparing to the methods of preliminary selection used before. Results of dissertational work were implemented in software for observation and detection of asteroids and comets in Andrushivka astronomical observatory, in Keldysh Institute of Applied Mathematics RAS and in software for processing of scanned astronomical photographic plates' in the Main Astronomical Observatory NAS of Ukraine. Using the present software there were about 900 asteroids and 2 comets' discoveries done in December 2011 and more than 200 thousands observations sent to the MPC. Scientific, theoretical and practical results of the present work can be used in: development of software for intraframe image processing for present and prospective optical observation facilities; education, training students in digital image processing while processing observations of novae and supernovae, optical afterglow of gamma-ray bursts.