An imaging spectrometer with a 256 element InGaAs diode array was combined with a high throughput optical arrangement for recording high quality NIR spectra (824 nm to 1700 nm) of plastics from a distance of 25 cm within 6.3 milliseconds. The considered spectral region was assessed to be suitable for plastic identification.
A spectroscopic near-infrared imaging system, using a focal plane array (FPA) detector, is presented for remote and on-line measurements on a macroscopic scale. On-line spectroscopic imaging requires high-speed sensors and short image processing steps. Therefore, the use of a focal plane array detector in combination with fast chemometric software is investigated. As these new spectroscopic imaging systems generate so much data, multivariate statistical techniques are needed to extract the important information from the multidimensional spectroscopic images. These techniques include principal component analysis (PCA) and linear discriminant analysis (LDA) for supervised classification of spectroscopic image data. Supervised classification is a tedious task in spectroscopic imaging, but a procedure is presented to facilitate this task and to provide more insight into and control over the composition of the datasets. The identification system is constructed, implemented, and tested for a real-world application of plastic identification in municipal solid waste.
An Adaptive Resonance Theory Based Artificial Neural Network (ART-2a) has been compared with Multilayer Feedforward Backpropagation of Error Neural Networks (MLF-BP) and with the SIMCA classifier. All three classifiers were applied to achieve rapid sorting of post-consumer plastics by remote near-infrared (NIR) spectroscopy. A new semiconductor diode array detector based on InGaAs technology has been experimentally tested for measuring the NIR spectra. It has been found by a cross validation scheme that MLF-BP networks show a slightly better discrimination power than ART-2a networks. Both types of artificial neural networks perform significantly better than the SIMCA method. A median sorting purity of better than 98% can be guaranteed for non-black plastics. More than 75 samples per second can be identified by the combination InGaAs diode array/neural network. However, MLF-BP neural networks can definitely not extrapolate. Uninterpretable predictions were observed in case of test samples that truly belong to a particular class but that are located outside the subspace defined by training set.
An infrared camera with focal plane InSb frared imaging technology for civil research as focal plane array delector has been applied lo the characterization of macroscopic samples of household waste over distances up to two meters. Per waste sample (singelized), a sequence of images was laken al six optical wavelength ranges in the near infrared region (1100 n m .. 2500 nm). The ob tained three-dimensional data stack served as individual array cameras containing rectangular arrangements of many thousands of miniaturised infrared sensitive delec tor elements. Distinct grades of these advanced detector materials not only cover ihe entire optical NIR wave length range (800-2500 nm), but also parts of the mid-in frared wavelength range. Additionally, these materials of fer the new possibility for rapid acquisition of at leasl fifty infrared images per second. In this way, these infrared cameras became attractive for real-time observations of chemical reactions and processes. Recently, Lewis, Levin and Treado |5, 6 , 7| reported first results with their InSb focal plane diode array for spectroscopic NIR-and MIRmicroscopy, Lodder [8 | reported a comparison of an InSb array versus a PtSi CCD array camera for medical NIR imaging, Marcott and Reeder (9| proposed (he use of NIRIS in a process-analytical industrial environment.A macroscopic application with a similar InSb camera type has been reported by us | 10, 11, 12, 13 1 for rapid pre sorting of household waste over a measuring distance of Near-and Mid-lnlrared Imaging Spectroscopy (NIRIS, Llp (0 two meters. InSb is optically sensitive between 1.1 f.tm MIR IS) are young and upcoming analytical techniques. iinci 5,5 ^nl which provides for simultaneous NIRIS and They form alternatives lo Raman imaging spectroscopy. MIR1S experiments. Chemometrieal methods were used fingerprint per sample. An abstract factor rotation of this stack of six images into a spectroscopieal meaningful in termediate six-element vector by Multivariate Image Rank Analysis (MIRA) finally provided a decision limit for the discrimination of plastics and nonplaslies. A correct clas sification of better than X0% has been reached. The ex perimental NIRIS set-up has been automated so far to al low an on-line identification of a real world waste sample within a few seconds.
IntroductionFuture expected specific advantages of NIRIS (resp. M1RIS) are high speed and high sensitivity. A lew years ago, Robert el al. 111, McClure |2, 31 and Geladi el al,|4 by the present authors to select the optimal subset of wavelengths and to decompose and to analyse the images. The aim of this work is to identify plastics in the waste starlccl lo use NIRIS for qualitative and quantitative image sircam by using the InSb local plane array as remote maanalysis of distinct organic and inorganic materials. They tcrial sensitive "image sensor. The enormous camera speed mainly used the optical near infrared short-wave region ant| Uie possibility to extract spectroscopic fingerprints of below 1700 nm. Mainly slow-scan videc...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.