NEAR-INFRARED IMAGING SPECTROSCOPY (NIRIS) AND IMAGE RANK ANALYSIS FOR REMOTE IDENTIFICATION OF PLASTICS IN MIXED WASTE

An infrared camera with focal plane InSb array detector has been appli ed to the characterization of macroscopic samples of household waste o ver distances up to two meters. Per waste sample (singelized), a seque nce of images was taken at six optical wavelength ranges in the near i nfrared region (1100 nm - 2500 nm). The obtained three-dimensional dat a stack served as individual fingerprint per sample. An abstract facto r rotation of this stack of six images into a spectroscopical meaningf ul intermediate six-element vector by Multivariate Image Rank Analysis (MIRA) finally provided a decision limit for the discrimination of pl astics and nonplastics. A correct classification of better than 80% ha s been reached. The experimental NIRIS set-up has been automated so fa r to allow an on-line identification of a real world waste sample with in a few seconds.