Multi-frequency laser radar band selection

Active, multi-spectral sensors are being investigated for the problem of detecting partially obscured, camouflaged and/or unresolved targets. The detection performance of these sensors depends upon numerous factors. The most significant of which are the number and location of the spectral bands, target to interference or clutter ratio, receiver SNR, and the level of speckle diversity. This paper addresses the problem of determining the optimal number and location of the spectral bands for a given target material and interference reflectance spectra. Three band selection algorithms are investigated; they are: 1) an interference-invariant version of the matched filter, 2) the maximum spectral angle, and 3) a new algorithm which is optimal for the posed problem but requires assumptions about the noise power level and fractional cover.