WASTE IDENTIFICATION AND SELECTION BY MEANS OF HIPERSPECTRAL NIR TECHNOLOGY

Recovered wood becomes more and more important for the production of wood-base panels. The increasing demand resulted in the need to improve the cleaning process of post-consumer material, eliminating in a more effective and efficient way plastic impurities. We have developed a new method that takes advantages of the NIR spectral images to identify different classes of materials present in wood waste. We have investigated the spectrum of a wide sample of materials as plastics, ceramics, tiles, woods, laminates in the range 1100 - 2500 nm. We found those features that characterized the different classes of materials and searched for those spectral regions able to distinguish them. We have studied the correlation among the various features that characterize the classes and the spectral bands potentially most effective in the discrimination process have been identified. We defined different indices able to distinguish among different materials. The developed classification method shows that the near infrared spectral analysis can be used as an efficient technique to identify different objects facilitating rapid and accurate separation process.