|Delwiche, Stephen - Steve|
Submitted to: Journal of Food Engineering
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/30/2013
Publication Date: 1/9/2014
Citation: Leewang-Hee, Kim, M.S., Lee, H., Delwiche, S.R., Bae, H., Kim, D., Cho, B. 2014. Hyperspectral near-infrared imaging for the detection of physical damages of pear. Journal of Food Engineering. 130:1-7. Interpretive Summary: Bruise damage on pear fruits is one of the most crucial internal quality factors that affects the value of the product. A reliable non-destructive detection method for bruises is needed to help ensure accurate fruit quality assessment. In this study, the feasibility of a novel near infrared hyperspectral imaging technique was investigated for the detection of bruise damages underneath pear skin. Pear bruises are not easily discernible by using conventional imaging technique in the visible wavelength ranges. The result demonstrated that pear bruises can be determined with an accuracy of 92%. This study illustrated that hyperspectral imaging technique using near-infrared spectral regions beyond human vision is well-suited for detection of below-surface fruit bruises. The methods and results presented is this study will benefit fruit growers and processors, as well as food technologists and engineers.
Technical Abstract: Bruise damage on pears is one of the most crucial internal quality factors, which needs to be detected in postharvest quality sorting processes. Thus, a reliable non-destructive detection method for the fruit defects including bruises is necessary to ensure accurate quality assessment. Infra-red imaging techniques (NIR) in the 1000 nm to 1700 nm have good potentials for identifying and detecting bruises since bruises result in the rupture of internal cell walls due to defects on agricultural materials. In this study, the feasibility of a novel NIR technique, hyperspectral imaging (1000 – 1700 nm), for the detection of bruise damages underneath the pear skin was investigated. Pear bruises, affecting the quality of fruits underneath the skin, are not easily discernible by using conventional imaging technique in the visible wavelength ranges. A classification algorithm based on F-value was applied for analysis of image in order to find the optimal waveband ratio for the discrimination of bruises against sound surface. The result demonstrated that the best threshold waveband ratio detected bruises with the accuracy of 92%. This study illustrated that the hyperspectral infra-red imaging technique with the region beyond NIR could be a potential detection method for pear bruises. Further studies on other classification algorithms and varieties of pears with variable storage time might empower the feasibility of hyperspectral image on the complete detection of defects on pears.