Artificial Intelligence Approach for Target Classification: A State of The Art

Abstract

The classification of static or mobile objects, from a signal or an image containing information as to their structure or their form, constitutes a constant concern of specialists in the electronic field. The remarkable progress made in past years, particularly in the development of neural networks and artificial intelligence systems, has further accentuated this trend. The fields of application and potential uses of Artificial Intelligence are increasingly diverse: understanding of natural language, visual recognition, robotics, autonomous system, Machine learning, etc. This paper is a state of the art on the classification of radar signals. It focuses on the contribution of artificial intelligence to the latter without forgetting target tracking. This by evoking the different feature extractors, classifiers and the existing identification deep learning algorithms. We detail also the process allowing carrying out this classification.

Automatic target detection and localization using ultra-wideband radar

“…The maximum correlation, identified as the maximum peak, indicates the delay (or time of flight) between transmitted and received pulses. This makes it possible to calculate the distance separating the target from the radar simply and reliably [28]. In our case, the calculated distance equal to 24 m.…”

“…Here the maximum correlation peak reflects the similarity between the two correlated signals, which implies the presence of the target. To test the ability of our algorithm to detect the real signals emitted by our detection radar and reflected by the target, we used an unknown signal [28], [29], and we correlated this signal with the template in order to test the similarity between them. Figures 5 and 6 show an unknown received signal and its correlation with the template.…”