OVERVIEW OF MODERN APPROACHES TO SOLVING THE PROBLEM OF EXTENDED RADAR TARGETS RECOGNITION

Abstract

The development of the theory and technology of radio location indicates the need to solve problems of radar recognition of targets by spatial parameters. This is due to the need to provide in promising radar stations (complexes) with a software overview of the required resolution to obtain three-dimensional images in all spatial coordinates.
 The task of radar targets recognition, which is to assign the observed objects to the appropriate classes and types, is of considerable and growing interest. Different classes (types) of targets make up a certain alphabet, the choice of which determines not only the effectiveness of the use of recognition, but also the difficulties that arise in its implementation.
 Currently, there are a large number of radar recognition algorithms. They differ in the stages of decision-making, the degree and nature of accounting for statistics of signs, obstacles and signals. Due to the fact that the secondary emission pattern of extended targets is multi-lobed, statistical algorithms are preferred.
 The information used for radar recognition is contained in the set of received radar signals. However, most often for target recognition certain measured target features are used, which are compared in accordance with the recognition algorithms with known (reference) features.
 The choice of recognition features is usually made heuristically. This set of parameters does not always allow providing the required quality of recognition. The synthesized algorithms work unstable or require unreasonably large computational costs due to a significant increase in the dimensionality of the feature space.
 Analysis of known radar recognition algorithms of extended targets shows that they were developed under significant constraints.
 At present, the tasks of point targets radar surveillance are most fully solved. The tasks of processing signals reflected from bodies of complex shape, given the difficulties of their formulation and solution are not fully explored.
 It is promising to consider a set of radar surveillance tasks and the criteria used in them as a multicriteria task, the solution of which is associated with vector optimization of the location system as a whole.