Spectral drift and correction technique for hyperspectral imaging systems

Abstract

A method for determining magnitude and direction of spectral channel drift for several consecutive spectral regions over a wide spectral range. According to the method of the present invention, in-field testing of a spectral filter sequentially irradiated by two blackbody sources is performed to generate a response function of the spectral filter. The response function is ensemble averaged to reduce any noise. Background radiance is then removed to produce a smoothed spectral transmittance function of the spectral filter. The first derivative function of the smoothed spectral transmittance function is determined. The first derivative function is separated into spectral band regions having +/− N pixels on either side of the function minima. The value of N is selected to optimize the detection algorithm sensitivity to change while extending the limit of spectral shift magnitude. The sum of the differences between the first derivative function and a reference spectral derivative function is determined. The difference result is applied to a look-up table to determine magnitude and direction of spectral drift for each of the separated spectral band regions. Use of the present invention can provide information on spectral distortion or spectral smile for 2-D focal plane arrays used for hyperspectral imaging.