Application of Hyperspectral Imaging for Surface Defects Detection of Jujube

Abstract

Hyperspectral imaging system with the range of 450-990 nm was developed to obtain the reflection spectral of Kaohsiung 11 jujube with surface defects. Principal component analysis (PCA) was used to reduce the spectral dimensionality of hyperspectral image data and determine the wavebands used by band ratio method for quick detection of jujube surface defects. Two-band ratio (Q550/680) images were successfully used to differentiate surfaces with defects such as decay, rusty, fungus infection and insect bites from the sound surface. Due to the fact that rusty surface of Kaohsiung 11 has no effect on the quality of flavor and texture, a threshold value for slope of reflectance spectra between 700 nm and 710 nm was used to differentiate the rusty region from other defect regions. The glare due to specular reflection from smooth and waxy surface of jujube may lead to error when the differentiation of surfaces with defects and sound surfaces was performed. The findings of this study can be used as a basis for developing effective algorithm to identify different types of defects on jujube surface.