Optical determination of material abundances by using neural networks for the derivation of spectral filters

摘要: Using appropriately designed spectral filters allows to optically determine material abundances. While an infinite number of possibilities exist for determining filters, we take advantage using neural networks derive leading precise estimations. To overcome some drawbacks that regularly influence the determination abundances hyperspectral data, incorporate variability raw materials into training considered networks. As a main result, successfully classify quantized optically. Thus, part high computational load, which belongs use networks, is avoided. In addition, derived become invariant against spatially varying illumination intensity as remarkable benefit in comparison with based on Moore-Penrose pseudoinverse, instance.