Detecting and Discriminating Small Arms Fire Using Neural Networks: Tools to Mitigate Urban Gun Violence

Abstract

This paper outlines a two-layer neural network used to declare small arms fire based upon a   temporal, electro-magnetic signature. The neural network discriminates targets from noise and common clutter sources, such as sun glints. This neural network is compared against a baseline algorithm derived from traditional detection and discrimination processing. A simulation is performed to evaluate the performance of each approach, and the simulation results are captured via a Receiver Operator Curve. This algorithm, in conjuction with a high-rate infrared sensor, would provide law-enforcement a tool for mitigating urban violence