Hyperspectral imaging techniques for detection of foreign materials from fresh-Cut vegetables

Authors: TUNNY, SALMA - Chungnam National University; KURNIAWAN, HARY - Chungnam National University; AMANAH, HANIM - Chungnam National University; BAEK, INSUCK - Orise Fellow; Kim, Moon; Chan, Diane; FARQEERZADA, MOHAMMAD - Chungnam National University; WAKHOLI, COLLINS - Chungnam National University; CHO, BYOUNG-KWAN - Chungnam National University

Submitted to: Postharvest Biology and Technology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/18/2023
Publication Date: 7/1/2023
Citation: Tunny, S., Kurniawan, H., Amanah, H., Baek, I., Kim, M.S., Chan, D.E., Farqeerzada, M., Wakholi, C., Cho, B. 2023. Hyperspectral imaging techniques for detection of foreign materials from fresh-Cut vegetables. Postharvest Biology and Technology. 201:112373. https://doi.org/10.1016/j.postharvbio.2023.112373.
DOI: https://doi.org/10.1016/j.postharvbio.2023.112373

Interpretive Summary: For fresh-cut vegetables sold as ready-to-eat or read-to-cook, the detection of any small foreign materials (FMs) in the product is an important part of food processing operations to ensure food safety and quality. This work developed models for hyperspectral imaging-based detection of FMs mixed into shredded or diced vegetables, specifically comparing the detection performances of models utilizing specially selected wavelengths of light for visible/near-infrared (VNIR) reflectance imaging, short-wave infrared (SWIR) imaging, and ultraviolet-excitation fluorescence imaging. Detection was tested for a wide variety of FMs, such as rubber, wood, paper, stone, metal, plastics, threads, cigarette butts, and dead insects, that were mixed into shredded or diced cabbage, carrot, green onion, onion, potato, radish, and zucchini. The detection model utilizing SWIR reflectance outperformed the models based on VNIR reflectance and on fluorescence. The results demonstrate that SWIR hyperspectral imaging may be an effective tool for non-destructive detection of foreign materials in fresh-cut product products that food processors could use to help ensure food safety and quality for their customers.

Technical Abstract: Foreign materials (FMs) in fresh-cut vegetables are a huge concern for the fresh-cut industry since they affect product safety and quality. Therefore, effective methods of detecting FMs in industrial processing operations are urgently required. In this study, three hyperspectral imaging (HSI) techniques (VNIR, SWIR, and fluorescence) were investigated to distinguish the FMs from seven common fresh-cut vegetables. In addition, a partial least squares discriminant analysis (PLS-DA) model was developed for all types of vegetables to identify the FMs. Among the three different HSI systems, SWIR provided the best FMs detection accuracy (99%), followed by VNIR (89%) and fluorescence (64%). Furthermore, the beta coefficient obtained from the PLS-DA model was applied to the hyperspectral image to reveal the FMs visually. Because the SWIR HSI system provided the best result, three variable selection techniques (SFS, SPA, and iPLS) were applied to select important wavelengths from the SWIR HSI data, and new PLS-DA models were developed. The results suggested that the SWIR HSI techniques with the SPA-PLS-DA model (99% overall detection accuracy) could be efficiently utilized in industrial applications for rapid and nondestructive FMs detection in fresh-cut vegetables.