Radar Plot Classification Based on Machine Learning

Abstract

Clutter is the inherent environment of radar signal detection and processing. On the one hand, too many clutter points will start a false track, on the other hand, clutter plots will be incorrectly associated with the track, resulting in track errors. Therefore, how to further distinguish target plots and clutter plots after target detection is an important and difficult problem in radar data processing. Aiming at the clutter data mixed in the plot data output by signal processing, this paper extracts the multi-dimensional feature parameters of radar plot on the radar measured data set, classifies the radar real target and false target by using the traditional support vector machine, fully connected neural network and convolution neural network respectively, and compares the effects of three different classification methods.