Researchers at the University of Nijmegen in Nijmegen, the Netherlands, have reported that preprocessing techniques can mitigate the undesirable scattering effects that compromise the infrared analysis of demolition waste. The group, which reported its results in Applied Spectroscopy, Vol. 55, No. 2, has developed a prototype sensor for the automated separation of wood, plastic and glass using near-infrared reflectance spectroscopy.

In experiments involving more than 1200 dry samples collected from two demolition sites in Italy and the Netherlands, the researchers investigated preprocessing methods that could yield higher-quality spectra. A modified standard normal variate technique enabled them to classify the samples using six regions in the 1154- to 1700-nm wavelength range.