An analog hardware detection scheme for passive remote infrared interferometers

This paper presents an analog detection scheme for very high speed signal processing of passive remote infrared interferograms. This system is comprised of three successive stages, all of which are performed in the analog domain. The first stage is an analog neural network which computes the Fourier transform of the interferogram. The second stage accepts as its inputs the real and imaginary parts of the Fourier transform and calculates the magnitude. The final stage is a neural network which uses the magnitude spectrum to determine the presence or absence of sulfur hexaflouride.<>