Description: The computing method of estimating the location of the fast moving object on a series of CCD images is developed. The method is based on a non-linear least-squares assessment of objects image parameters. The minimization of sum squares of the residuals is solved by the Levenberg-Markvard method. The brightness of pixels of the each hypothetical object image in SIFP is distributed according to a mix of normal laws. Using this method the approximation of extended objects image by the Gaussian mixture models provides sufficient accuracy of the location of extended objects. The method can be used in programs of the automated detection of small solar system bodies when assessing the fast moving objects location, such as asteroids approaching the Earth.
Keywords: processing of CCD images, stretched object, assessment of location, CoLiTec