|MO, CHANGYEUN - Korean Rural Development Administration|
|KIM, GIYOUNG - Korean Rural Development Administration|
|LIM, JONGGUK - Korean Rural Development Administration|
|LEE, SEUNG - Chungnam National University|
|LEE, HONGSEK - Korean Rural Development Administration|
|KANG, JONGSUK - Korean Rural Development Administration|
|CHO, BYOUNG-KWAN - Chungnam National University|
Submitted to: Infrared Physics and Technology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/7/2017
Publication Date: 5/9/2017
Citation: Mo, C., Kim, G., Kim, M.S., Lim, J., Lee, S.H., Lee, H., Kang, J., Cho, B. 2017. Discrimination methods of biological contamination on fresh-cut lettuce based on VNIR and NIR hyperspectral imaging. Infrared Physics and Technology. 85:1-12.
Interpretive Summary: Foreign organisms and inanimate materials, such as worms, slugs, insects, gravel, and soil are often included during harvesting of lettuce. These contaminants can affect the safety and quality of the lettuce. The potential use of noninvasive imaging methods that can be used on automated processing lines for fresh produce, to screen out such foreign objects, is of great interest for improving food safety and processing efficiencies. In this study, multispectral image-based algorithms using visible and near-infrared light were developed to detect worms on fresh-cut lettuce and demonstrated 97% detection accuracy in a laboratory imaging environment. The image-based algorithms can be implemented for rapid online screening of worms on fresh produce in a processing line. Further research is planned to develop additional algorithms that can be used to simultaneously detect other foreign items as well. The techniques developed in this investigation will benefit fresh produce producers and processors by providing methods to sort produce contaminated with foreign materials.
Technical Abstract: Multispectral imaging algorithms were developed using visible-near-infrared (VNIR) and near-infrared (NIR) hyperspectral imaging (HSI) techniques to detect worms on fresh-cut lettuce. The optimal wavebands that detect worm on fresh-cut lettuce for each type of HSI were investigated using the one-way ANOVA analysis. The worm detection imaging algorithms for the VNIR and NIR imaging resulted in a prediction accuracy of 97.00% for RI547/945, and 100.0% for RI1064/1176, SI1064-1176, RSI-I(1064-1173)/1064, and RSI-II(1064-1176)/(1064+1176), respectively. The two HSI techniques showed that the spectral images with a pixel size of 1 × 1 mm or 2 × 2 mm had the best classification accuracy for worms. The overall results demonstrate that hyperspectral reflectance imaging techniques have the potential to detect worm on fresh-cut lettuce. In the future, our research will focus on a real-time sorting system for lettuce to simultaneously detect various defects such as browning, worms, and slugs.