Noise, clutter, and determinism in infrared spectral signatures under varying meteorological conditions

Abstract

Infrared remote detection of chemical and biological agents in a complex environment depends on the ability to recognize threat signatures and differentiate them from the signatures of innocuous materials. In this paper, we addressed the methods of producing the constraint spectra needed to ensure reliable operation in a meteorologically changing environment. We collected arrays of background spectra of ground, woods, and low-angle sky on an irregular basis over a period of a year. Based on the hypothesis that the concentration fluctuations of species in the sensor's field of view can be exploited to form signatures, the standard deviations of the array (the result is characteristic of all fluctuations) and the difference array (the result is characteristic of sensor fluctuations) were computed. Subtracting these two spectra and filtering the result produced a spectrum, which is a measure of the IR fluctuations in the scene. The resulting set of scene spectra were processed into aberrant noise, and deterministic groups by numerical filtering and statistical methods.