Assessment of the optical response of bruised kiwifruit using hyperspectral imaging and its relationships with water migration

Abstract

This study aimed to investigate the capability of hyperspectral imaging (HSI) technology in non-destructive detection of early damage in kiwifruit, revealing its detection mechanisms from the perspective of moisture migration caused by damage. Low-field nuclear magnetic resonance (LF NMR) technology, HSI, and hyperspectral microscopic imaging (HMI) were employed to examine the relationships between moisture content in different states, relative spectral reflectance, and spectral absorbance of “Zespri” green kiwifruit at 0, 2, 4, 6, and 8 d post-damage. Additionally, the relationship between spectral absorbance and moisture content in four states was explored. The results indicated that bruising damage disrupted the microstructure of the kiwifruit, and exacerbated the migration of free water in the vacuoles, leading to an earlier peak in the proportion of free water compared to the control group. Significant moisture absorption bands were observed at the characteristic wavelengths corresponding to hyperspectral relative reflectance and hyperspectral microscopic absorbance of damaged kiwifruit. A high negative correlation was found between hyperspectral microscopic absorbance and free water content, with linear fitting results for the top five wavelengths showing R² values exceeding 0.8940. Using the hyperspectral reflectance at the 30 most relevant characteristic wavelengths related to free water content, early discrimination analysis of damaged kiwifruit was conducted, yielding an average accuracy of 96.56 % across four model validation sets. These findings indicate that the mechanism of hyperspectral imaging in detecting bruises in kiwifruit is potentially related to free water content. This research provides new insights into the non-destructive detection mechanisms of bruised fruit using hyperspectral imaging technology from the perspective of moisture migration.