Decomposition of the scattered field into singularities for object classification using artificial intelligence algorithms

Yasmina Zaky

One of the applications of radar is the detection and identification of objects from their ultra-wideband scattered field response. Indeed, work of this type was initiated by C.E. Baum who proposed to apply the singularity expansion methods (SEM) to the scattered field of an object illuminated by a broadband incident wave. The extraction and study of the resonant poles of these measured signals allows to distinguish different objects by identifying their natural poles. In this thesis, the SEM is explored in order to establish a compact model that accurately represents the ultra-wideband scattered field of an object independently of the observation angle and its orientation. In this perspective, several SEM techniques were compared: TLS Matrix Pencil for time domain signals, TLS Cauchy, and Vector Fitting for frequency domain signals.

Following the frequency discrimination of objects obtained by the SEM technique, supervised classification algorithms of the Machine Learning and Deep Learning type are applied to classify different objects from their characteristic parameters. Hence, several classification algorithms have been studied: Support Vector Machine (SVM), Decision Tree (DT), Multi-Layer Perceptron (MLP) and Convolutional Neural Network (CNN). This study shows that the combination of a noise-robust SEM technique with neural network-based classifiers allows to classify the shape or the material of an object from a single measurement and with a low computational cost. Moreover, we propose a procedure that allows to determine the direction of the receiving antenna and the orientation of an object from the residues that are associated with each resonant pole. This classification procedure using data from the SEM is very promising especially when generalizing to data not included in the training set.

I received my PhD in electronics from the University of Côte d’Azur, Nice, France in 2022. I am currently working as a research engineer in electromagnetism/antennas. My research interests include radiation and diffraction measurements, miniature antennas, machine learning and deep learning.