Microplastic abundance quantification via a computer-vision-based chemometrics-assisted approach
Bertoldi, Crislaine LU
; Lara, Larissa Z.
; Gomes, Adriano A.
and Fernandes, Andreia N.
(2021)
In Microchemical Journal
160, Part B.
- Abstract
Microplastic (MP) contamination is a topic of growing global concern; these particles are ubiquitous in environmental ecosystems and have been found in aquatic, terrestrial, and atmospheric mediums. However, the protocols to quantify MPs in environmental samples have limitations and may lead to overestimation and/or underestimation of the plastic debris. Therefore, the aim of this research was to develop a simple procedure to determine the abundance of MPs using digital image processing and chemometric treatment. The proposed method combined computer-vision-based and multivariate calibration by partial least squares coupled with interval selection (iPLS and successive algorithm projection - iSPA). The abundance ranges of the yellow,... (More)
Microplastic (MP) contamination is a topic of growing global concern; these particles are ubiquitous in environmental ecosystems and have been found in aquatic, terrestrial, and atmospheric mediums. However, the protocols to quantify MPs in environmental samples have limitations and may lead to overestimation and/or underestimation of the plastic debris. Therefore, the aim of this research was to develop a simple procedure to determine the abundance of MPs using digital image processing and chemometric treatment. The proposed method combined computer-vision-based and multivariate calibration by partial least squares coupled with interval selection (iPLS and successive algorithm projection - iSPA). The abundance ranges of the yellow, blue, black, colourless, green, and red MPs were 1–212, 7–134, 0–50, 6–290, 0–113, and 20–392, respectively. When the models were applied to an independent set of samples, the following RMSEP values were found: 9.8 (yellow), 6.4 (blue), 3.5 (black), 8.1 (colourless), 7.5 (green), and 19.3 (red). The results showed that image processing has the potential to quantify MPs with respect their colour. This method could help to reduce time-consuming and to avoid subjectivity in future analyses.
(Less)
- author
- Bertoldi, Crislaine
LU
; Lara, Larissa Z.
; Gomes, Adriano A.
and Fernandes, Andreia N.
- publishing date
- 2021-01
- type
- Contribution to journal
- publication status
- published
- keywords
- Colour histogram, Image processing, Microplastic contamination, Multivariate calibration
- in
- Microchemical Journal
- volume
- 160, Part B
- article number
- 105690
- publisher
- Elsevier
- external identifiers
-
- scopus:85096185092
- ISSN
- 0026-265X
- DOI
- 10.1016/j.microc.2020.105690
- language
- English
- LU publication?
- no
- additional info
- Publisher Copyright: © 2020
- id
- f038e76d-be16-4b20-938f-90b7a31a1b88
- date added to LUP
- 2024-07-02 09:12:00
- date last changed
- 2024-07-03 12:31:21
@article{f038e76d-be16-4b20-938f-90b7a31a1b88,
abstract = {{<p>Microplastic (MP) contamination is a topic of growing global concern; these particles are ubiquitous in environmental ecosystems and have been found in aquatic, terrestrial, and atmospheric mediums. However, the protocols to quantify MPs in environmental samples have limitations and may lead to overestimation and/or underestimation of the plastic debris. Therefore, the aim of this research was to develop a simple procedure to determine the abundance of MPs using digital image processing and chemometric treatment. The proposed method combined computer-vision-based and multivariate calibration by partial least squares coupled with interval selection (iPLS and successive algorithm projection - iSPA). The abundance ranges of the yellow, blue, black, colourless, green, and red MPs were 1–212, 7–134, 0–50, 6–290, 0–113, and 20–392, respectively. When the models were applied to an independent set of samples, the following RMSEP values were found: 9.8 (yellow), 6.4 (blue), 3.5 (black), 8.1 (colourless), 7.5 (green), and 19.3 (red). The results showed that image processing has the potential to quantify MPs with respect their colour. This method could help to reduce time-consuming and to avoid subjectivity in future analyses.</p>}},
author = {{Bertoldi, Crislaine and Lara, Larissa Z. and Gomes, Adriano A. and Fernandes, Andreia N.}},
issn = {{0026-265X}},
keywords = {{Colour histogram; Image processing; Microplastic contamination; Multivariate calibration}},
language = {{eng}},
publisher = {{Elsevier}},
series = {{Microchemical Journal}},
title = {{Microplastic abundance quantification via a computer-vision-based chemometrics-assisted approach}},
url = {{http://dx.doi.org/10.1016/j.microc.2020.105690}},
doi = {{10.1016/j.microc.2020.105690}},
volume = {{160, Part B}},
year = {{2021}},
}