Hyperspectral imaging (HSI) is currently more and more utilized in waste recycling industry for both sensor-based sorting and quality control applications. Plastic waste is one of the flow streams in which HSI is particularly effective, due to its high capability of polymer identification in the near and short-wave infrared range, allowing to achieve high purity recycled plastic products and, therefore, secondary raw materials characterized by high quality. The aim of this work was to evaluate the potential of HSI-based data fusion, to achieve simultaneous identification of post-consumer plastic packaging flakes by polymer and color. Five different polymers among those commonly used for plastic packaging, i.e., polystyrene (PS), polyethylene terephthalate (PET), expanded polystyrene (EPS), polyethylene (PE), and polypropylene (PP), subdivided into 6 different color classes (orange, red, transparent, green, blue and white) were investigated. Two different HSI devices were used to perform the polymer and color identification, operating in the short-wave infrared range (1000-2500 nm) and in the visible range (400-750 nm), respectively. A hierarchical classification model based on partial least square - discriminant analysis (PLS-DA) was built in order to obtain a high-level efficiency in prediction for all classes. The performances of the model were evaluated in terms of sensitivity, specificity, precision and F1 score. The obtained results were very promising, showing how HSI coupled with data fusion can be utilized as a non-invasive, fast and efficient tool to obtain high-quality recycled plastics, optimizing the industrial plastic recycling process.
Hyperspectral imaging coupled with data fusion for plastic packaging waste recycling / Bonifazi, Giuseppe; Capobianco, Giuseppe; Cucuzza, Paola; Serranti, Silvia. - 12327:(2023), pp. 1-14. (Intervento presentato al convegno SPIE Future Sensing Technologies 2023 tenutosi a Yokohama, Japan) [10.1117/12.2645119].
Hyperspectral imaging coupled with data fusion for plastic packaging waste recycling
Giuseppe BonifaziPrimo
Supervision
;Giuseppe CapobiancoSecondo
Methodology
;Paola CucuzzaPenultimo
Investigation
;Silvia Serranti
Ultimo
Conceptualization
2023
Abstract
Hyperspectral imaging (HSI) is currently more and more utilized in waste recycling industry for both sensor-based sorting and quality control applications. Plastic waste is one of the flow streams in which HSI is particularly effective, due to its high capability of polymer identification in the near and short-wave infrared range, allowing to achieve high purity recycled plastic products and, therefore, secondary raw materials characterized by high quality. The aim of this work was to evaluate the potential of HSI-based data fusion, to achieve simultaneous identification of post-consumer plastic packaging flakes by polymer and color. Five different polymers among those commonly used for plastic packaging, i.e., polystyrene (PS), polyethylene terephthalate (PET), expanded polystyrene (EPS), polyethylene (PE), and polypropylene (PP), subdivided into 6 different color classes (orange, red, transparent, green, blue and white) were investigated. Two different HSI devices were used to perform the polymer and color identification, operating in the short-wave infrared range (1000-2500 nm) and in the visible range (400-750 nm), respectively. A hierarchical classification model based on partial least square - discriminant analysis (PLS-DA) was built in order to obtain a high-level efficiency in prediction for all classes. The performances of the model were evaluated in terms of sensitivity, specificity, precision and F1 score. The obtained results were very promising, showing how HSI coupled with data fusion can be utilized as a non-invasive, fast and efficient tool to obtain high-quality recycled plastics, optimizing the industrial plastic recycling process.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.