Method of space debris element recognition from optical and radar images based on neural network technologies

Abstract

An improved Bayesian method is presented for recognising space debris elements from optical and radar images and is intended to identify potentially hazardous space debris elements for operational spacecraft. A set of informative features for recognising space debris elements is proposed. A modified Bayesian classifier based on a deep neural network with a sequential decision-making procedure is considered as a decisive rule. The quality requirements for the used optical and radar images are defined. The results of recognition probability estimation for different types of space debris elements with different number of used optical and radar images are obtained.