Matched filter for extended images of astronomical objects in CCD-frames from streams

In this paper we presented the developed matched filter and its processing results for the extended images of objects in series of CCD-frames that were fed in an online mode in the form of data streams and taken without the diurnal tracking. This matched filter has a non-analytical profile and is incapsulated into the CoLiTecSAT software. It was designed for machine vision purpose with selection the objects images of known shape against the background noise based on the pattern recognition techniques. Parameters of the matched filter are determined automatically by the results of frame pre-processing and are unique for each frame. In this case, it allows improving the image segmentation of geostationary satellites and reference stars. Also, this approach allowed reducing the number of false detections, increasing the accuracy of measuring both reference stars in the frame and the investigated satellite and, accordingly, increasing the accuracy of calculation and prediction of its orbit.