Optimizing plastic waste sorting with filter-enhanced NIR imaging: A multi-stage sensor-based approach for material recycling facilities
Efficient recovery of mixed post-consumer plastics requires both material recognition and reliable physical separation. This study evaluates a filter-assisted near-infrared (NIR) imaging workflow using packaging waste from an operating single-stream material recovery facility. Particle images were acquired with an InGaAs line-scan camera under three aperture settings (f/8, f/11, and f/16), using unfiltered and long-pass-filtered conditions with nominal cut-on wavelengths of 1100, 1300, and 1500 nm. Fixed-range normalization produced dimensionless detector responses, rather than calibrated reflectance. Principal component analysis (PCA) was used to examine response patterns, and empirical single-filter response thresholds were applied in a sequential laboratory sorting scheme. Reported recoveries were approximately 90% for PP, 99.9% for PS, 61% for HDPE, 50% for LDPE, and 94% for PET; LDPE was obtained as the residual fraction. These results describe technical verification on the studied material, because characterized particles were reused and unidentified items were provisionally assigned through NIR/PCA. They do not establish independent predictive accuracy. The contribution is the experimental integration of physical long-pass filtering with an interpretable, staged sorting workflow. Calibrated acquisition and independent, chemically characterized test samples are required to assess transferability.
Authors
- Ergin Gülcan (ORCID: https://orcid.org/0000-0002-8861-8061)
Institutions
- Hacettepe University (TR)
Publication Details
- Journal
- Separation Science and Technology
- Published
- 2026-11-21
- DOI
- https://doi.org/10.1080/01496395.2026.2767121
- Primary Topic
- Spectroscopy and Chemometric Analyses
- Type
- article
- Field-Weighted Citation Impact
- 0.00