Abstract

Along with the improvement of radar technologies, Automatic TargetRecognition (ATR) using Synthetic Aperture Radar (SAR) and Inverse SAR (ISAR)has come to be an active research area. SAR/ISAR are radar techniques togenerate a two-dimensional high-resolution image of a target. Unlike othersimilar experiments using Convolutional Neural Networks (CNN) to solve thisproblem, we utilize an unusual approach that leads to better performance andfaster training times. Our CNN uses complex values generated by a simulation totrain the network; additionally, we utilize a multi-radar approach to increasethe accuracy of the training and testing processes, thus resulting in higheraccuracies than the other papers working on SAR/ISAR ATR. We generated ourdataset with 7 different aircraft models with a radar simulator we developedcalled RadarPixel; it is a Windows GUI program implemented using Matlab andJava programming, the simulator is capable of accurately replicating a realSAR/ISAR configurations. Our objective is to utilize our multi-radar techniqueand determine the optimal number of radars needed to detect and classifytargets.

 

Introduction (beta)

None

 

Conclusion (beta)

None