Sieriebriakov A. Methods for object type identification by audio and radar signals

In the thesis, the studying of audio and radar signals was carried out, their regularities and methods of processing were determined. Methods of identifying types of objects by audio and radar signals have been developed by creating databases of reference parameters and comparing them with the parameters of the input signals. The evaluation criteria, according to which this comparison should be carried out, have been determined. Based on criteria, the use of developed information technologies is proposed. The expediency of using object identification methods in audio signals to determine the types of sources during acoustic monitoring of UAV-type flying objects in the surrounding air environment, audio classification of sources, audio coding or sound diagnostics of technical equipment is substantiated; in radar signals to determine the types of naval ships and ensure safe sea and river navigation. Object of research: processes of recognition of types of sources of audio signals and naval surface ships types by reflected radar signals. Subject of research: methods for detection and identification of object types in signals with high-frequency noise, information technologies, features and criteria for classification of objects by multicomponent audio signals and by reflected radar signals for determining audio sources type and naval ships type. The purpose of research is providing effective objects recognition by audio and radar signals with noise by developing and using methods for identification of aerial object type by audio signals and sea ship type by reflected radar signals. In the first chapter an overview and results of analysis of existing methods of object identification in complex signals of different physical nature are given, obtained by various means. Methods of analysis and separation of audio signal sources are considered. An overview and analysis of the most used methods of processing audio and radar signals for solving the problems of recognition, classification of audio sources and sea ships was carried out. Characteristics of multicomponent signals and methods of their processing are presented. The second section describes the method of identifying the source of an audio signal by the signature of characteristic peaks of salient frequencies. The characteristics of the use of time-frequency separation of components are given. Methods and algorithms for implementing such audio signal processing are shown. The process of obtaining the frequency signature of the source is characterized. The criterion for determining the similarity of the input signal to the signature is given. In the third section, a histogram-normalized method of identifying the type of sea ship based on reflected radar signals is provided. The principles of radar recognition of object types are substantiated. A recognition algorithm and a method of forming a database of reference parameters of various types of naval ship objects are proposed. An identification criterion was developed based on the characteristics of the comparison of the input signal with the reference parameters. The fourth chapter is devoted to the practical application of the developed methods of object identification in information technologies, the method of applying the specified criteria for the identification of object types is characterized. The methods of detecting the source of audio signals of unmanned aerial vehicles (UAVs) during sound aerial monitoring are disclosed, the characteristics of sound noises and distrubances, methods of filtering and analysis of such audio signals, methods of identifying types of objects are given. Quality assessments and results of identification testing of various types of naval ships are given.