PRI Deinterleaving with Connected Component Labeling-Based Clustering

Abstract

Air platforms use radar warning receiver systems to detect threat radars in military operations. It is critical for an aircraft operating in a hostile environment to detect and classify radar signals and determine radar emitters' identity and capability. This study proposes an innovative solution to the signal separation problem for pulse radar signals by the computer vision-based connected component labeling method. With the proposed solution, clustering is performed automatically on the three-dimensional image matrix created by using the parameters of the arrival angle, radio frequency, pulse width of the signals. After clustering, the pulse repetition interval (PRI) deinterleaving is provided by the time of arrival analysis in each cluster. Also, various simulations were carried out using different synthetic radar datasets containing pulse-on-pulse states. In the simulations, PRI accuracy analyzes were performed.