Uss M. Parameter estimation and filtering of remote sensing images based on their selfsimilar (fractal) structure

The dissertation is devoted to the questions of quality improvement of remote sensing (RS) images filtering and interpretation. It is proposed to solve this questions by means of development of methods and algorithms of selfsimilar (fractal) fields parameters estimation and filtering in the presence of noise with unknown parameters. Scientific results are: 1) a method of filtering images with selfsimilar (fractal) structure by joint processing of similar image fragments formed on different scales, which allows to use redundancy connected with their selfsimilar structure to improve filtering quality; 2) principles of parameters estimation and filtering of fractal Brownian motion field in the presence of additive, multiplicative and impulsive noise, which allow to take into account presence of edges on images; 3) principles of noise parameters estimation for images with selfsimilar (fractal) structure: additive and multiplicative noise variance and impulse noise corruption probability. The developed methods and principles allow to improve quality of remote sensing (RS) images filtering, quality of estimation of parameters which characterize their selfsimilar (fractal) structure and quality of estimation image noise parameters.