Title :
Evaluation of time-gated fluorescence spectroscopy for the classification of fluorescently labeled plastics
Author :
Fomin, Petr ; Brunner, Siegfried ; Kargel, Christian
Author_Institution :
Div. of Sensor Technol. & Meas. Syst., Bundeswehr Univ. Munich, Neubiberg, Germany
Abstract :
The automated recycling of plastics becomes increasingly attractive from both the economical and environmental point of view. Yet in applications where plastic parts with high quality are required, recycled plastic materials cannot be employed at present because waste plastics of different kind are usually mixed during the recycling process. In order to produce the necessary mono-fractional compositions of waste plastics, a highly reliable sorting method before recycling is of paramount importance. We have been investigating the promising approach of labeling virgin plastics with specifically designed fluorescent tracers (“markers”) in earlier studies [1]-[3]. Despite all the advantages of this approach compared to other techniques, the overlap of the tracers´ fluorescence with the intrinsic autofluorescence plastic polymers might show is quite challenging in terms of a highly reliable classification. In this study we compare steady-state spectroscopy so far implemented in our sorting system prototype with time-gated fluorescence spectroscopy based upon their classification performances. As it turns out, the steady-state approach delivers acceptable performance only at rather high signal-to-noise ratios (S/N) when pronounced autofluorescence is present. Such S/N levels might be hard to achieve in practice. The time-gated approach on the other hand can completely eliminate the intrinsic autofluorescence from the measured spectra at the cost of moderately decreasing the signal-to-noise ratio. As long as these S/N values are larger than about +10 dB, ideal classification results of F1-score = 100% can be achieved in practice for the application at hand.
Keywords :
fluorescence spectroscopy; image classification; industrial waste; plastics; production engineering computing; recycling; S/N levels; autofluorescence plastic polymers; automated recycling; fluorescent tracers; fluorescently labeled plastics classification; intrinsic autofluorescence; mono-fractional compositions; recycled plastic materials; signal-to-noise ratio; signal-to-noise ratios; steady-state spectroscopy; time-gated approach; time-gated fluorescence spectroscopy; waste plastics; Noise; Photonics; Plastics; Sorting; Spectroscopy; Steady-state; Wavelength measurement;
Conference_Titel :
Instrumentation and Measurement Technology Conference (I2MTC), 2015 IEEE International
Conference_Location :
Pisa
DOI :
10.1109/I2MTC.2015.7151357