Relevance and challenge
The reliable identification and separation of plastics is a key component of modern recycling systems. PVC (polyvinyl chloride) presents a particular challenge due to its spectral variability, influenced by pigments, additives, thermal aging, and origin. In mixed fractions containing PE, PP, and other plastics, material-specific detection becomes technically demanding.
As part of its strategic initiative to explore advanced sorting technologies, ROVI-TECH (Belgium) partnered with Cubert GmbH to assess the industrial applicability of hyperspectral imaging for accurate PVC separation in realistic sorting scenarios.
Research approach – Methodological procedure in the practical test
Initiated by ROVI-TECH (Belgium), the study evaluated hyperspectral sorting technologies under industry-relevant conditions. Two camera systems from Cubert were used:
- ULTRIS XMR (visible and near-infrared spectrum – VNIR)
- ULTRIS SWIR 1 (short-wave infrared – SWIR)
Heterogeneous plastic samples of varying colors, origins, and chemical compositions were tested. . Machine learning models were trained and validated using previously unseen samples.
Spectral truth under industrial conditions
Analyses with the ULTRIS XMR revealed significant intra-group variability among PVC, PE, and PP, leading to overlapping spectral signatures in the VNIR range (Fig. 1). While controlled tests yielded precise classifications, accuracy dropped under realistic conditions: 81% overall accuracy, but only 62% for PVC (Fig. 2).
Supplementary physical methods like the float/sink test helped correct overclassifications, but false negatives for PVC remained a challenge.
In contrast, the ULTRIS SWIR 1 camera provided consistently reliable PVC detection, even with varied colors and surface features (Fig. 3). Validation showed 81% overall accuracy and 100% PVC detection. Misclassifications were mitigated using physical separation techniques (Fig. 4).
Hyperspectral analysis as a robust tool for industrial sorting processes
The joint study confirms that SWIR-based hyperspectral imaging, combined with physical separation methods, offers an effective solution for sorting PVC in mixed plastic streams. In industrial environments with high material diversity, SWIR analysis outperforms VNIR approaches.
This advantage stems from the dominance of color in the VNIR spectrum and the unique material characteristics detectable only in the SWIR range. Hyperspectral technologies show strong potential when integrated into data-driven material recovery processes. Combining both camera types may unlock further insights for future applications.
Typical application scenarios
- Automated sorting of PVC in post-consumer and industrial waste
- Incoming goods inspection for contamination detection
- In-process quality monitoring during manufacturing
Rovi-Tech (Belgium) has been designing control, inspection, measurement, and sorting systems and machines for over 30 years to enhance industrial performance. By integrating industrial vision technologies and advanced measuring instruments, Rovi-Tech develops automated equipment ranging from feasibility studies to prototype development and the implementation of industrial installations. The techniques used include cameras, laser triangulation methods, infrared thermographic cameras, hyperspectral cameras, among others.
About the Author
Dr. Matthias Locherer has been the Sales Director at Cubert GmbH since 2017. With a PhD in Earth Observation from Ludwig Maximilian University of Munich, he brings extensive expertise in remote sensing, spectral imaging, and data analysis. Matthias has contributed to numerous research projects and publications, particularly in the hyperspectral monitoring of biophysical and biochemical parameters using hyperspectral satellite missions. His deep knowledge of optical measurement techniques and physical modeling makes him a key driver in advancing innovative hyperspectral technologies at Cubert.
