Plastic classification via in-line hyperspectral camera analysis and unsupervised machine learning

Martin L. Henriksen; Celine B. Karlsen; Pernille Klarskov; Mogens Hinge

doi:10.1016/j.vibspec.2021.103329

Plastic classification via in-line hyperspectral camera analysis and unsupervised machine learning

Martin L. Henriksen, Celine B. Karlsen, Pernille Klarskov, Mogens Hinge

Department of Biological and Chemical Engineering - Process and Materials Engineering
Department of Electrical and Computer Engineering - Electronics and Photonics

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaper › Journal article › Research › peer-review

77 Downloads (Pure)

Abstract

An increase in the quality of recycled plastic is paramount to address the global plastic challenge and applicability of recycled plastics. A potent approach is mechanical plastic sorting but sufficient analytical techniques are needed. This study applies unsupervised machine learning on short wave infrared hyperspectral data to build a model for classification of plastics. The model can successfully distinguish between twelve plastics (PE, PP, PET, PS, PVC, PVDF, POM, PEEK, ABS, PMMA, PC, and PA12) and the utility is further proven by recognizing three unknown samples (PS, PMMA, PC). The experimental setup is constructed similar to an in-line industrial setup, and the machine learning is optimized for minimal data processing. This ensures the industrial relevance and is a stepping-stone to solve the global plastic challenge.

Original language	English
Article number	103329
Journal	Vibrational Spectroscopy
Volume	118
ISSN	0924-2031
DOIs	https://doi.org/10.1016/j.vibspec.2021.103329
Publication status	Published - Jan 2022

Access to Document

10.1016/j.vibspec.2021.103329

1-s2.0-S0924203121001247-mainFinal published version, 1.68 MBLicence: CC BY

https://www.sciencedirect.com/science/article/pii/S0924203121001247

Cite this

@article{13cd4b45539c40149122bed85417cf1b,

title = "Plastic classification via in-line hyperspectral camera analysis and unsupervised machine learning",

abstract = "An increase in the quality of recycled plastic is paramount to address the global plastic challenge and applicability of recycled plastics. A potent approach is mechanical plastic sorting but sufficient analytical techniques are needed. This study applies unsupervised machine learning on short wave infrared hyperspectral data to build a model for classification of plastics. The model can successfully distinguish between twelve plastics (PE, PP, PET, PS, PVC, PVDF, POM, PEEK, ABS, PMMA, PC, and PA12) and the utility is further proven by recognizing three unknown samples (PS, PMMA, PC). The experimental setup is constructed similar to an in-line industrial setup, and the machine learning is optimized for minimal data processing. This ensures the industrial relevance and is a stepping-stone to solve the global plastic challenge.",

keywords = "Plastic identification, Hyperspectral imaging, Plastic recycling, Principal component analysis",

author = "Henriksen, {Martin L.} and Karlsen, {Celine B.} and Pernille Klarskov and Mogens Hinge",

year = "2022",

month = jan,

doi = "10.1016/j.vibspec.2021.103329",

language = "English",

volume = "118",

journal = "Vibrational Spectroscopy",

issn = "0924-2031",

publisher = "Elsevier",

}

Plastic classification via in-line hyperspectral camera analysis and unsupervised machine learning. / Henriksen, Martin L.; Karlsen, Celine B.; Klarskov, Pernille et al.
In: Vibrational Spectroscopy, Vol. 118, 103329, 01.2022.

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaper › Journal article › Research › peer-review

TY - JOUR

T1 - Plastic classification via in-line hyperspectral camera analysis and unsupervised machine learning

AU - Henriksen, Martin L.

AU - Karlsen, Celine B.

AU - Klarskov, Pernille

AU - Hinge, Mogens

PY - 2022/1

Y1 - 2022/1

N2 - An increase in the quality of recycled plastic is paramount to address the global plastic challenge and applicability of recycled plastics. A potent approach is mechanical plastic sorting but sufficient analytical techniques are needed. This study applies unsupervised machine learning on short wave infrared hyperspectral data to build a model for classification of plastics. The model can successfully distinguish between twelve plastics (PE, PP, PET, PS, PVC, PVDF, POM, PEEK, ABS, PMMA, PC, and PA12) and the utility is further proven by recognizing three unknown samples (PS, PMMA, PC). The experimental setup is constructed similar to an in-line industrial setup, and the machine learning is optimized for minimal data processing. This ensures the industrial relevance and is a stepping-stone to solve the global plastic challenge.

AB - An increase in the quality of recycled plastic is paramount to address the global plastic challenge and applicability of recycled plastics. A potent approach is mechanical plastic sorting but sufficient analytical techniques are needed. This study applies unsupervised machine learning on short wave infrared hyperspectral data to build a model for classification of plastics. The model can successfully distinguish between twelve plastics (PE, PP, PET, PS, PVC, PVDF, POM, PEEK, ABS, PMMA, PC, and PA12) and the utility is further proven by recognizing three unknown samples (PS, PMMA, PC). The experimental setup is constructed similar to an in-line industrial setup, and the machine learning is optimized for minimal data processing. This ensures the industrial relevance and is a stepping-stone to solve the global plastic challenge.

KW - Plastic identification

KW - Hyperspectral imaging

KW - Plastic recycling

KW - Principal component analysis

U2 - 10.1016/j.vibspec.2021.103329

DO - 10.1016/j.vibspec.2021.103329

M3 - Journal article

SN - 0924-2031

VL - 118

JO - Vibrational Spectroscopy

JF - Vibrational Spectroscopy

M1 - 103329

ER -