MULTISPECTRAL FLUORESCENCE IMAGING TECHNIQUE FOR DISCRIMINATION OF CUCUMBER SEED VIABILITY

摘要

In this study, we developed a nondestructive method for discriminating viable cucumber (Cucumis sativus) seeds based on hyperspectral fluorescence imaging. The fluorescence spectra of cucumber seeds in the 425-700 nm range were extracted from hyperspectral fluorescence images obtained using 365 nm ultraviolet excitation. Four kinds of algorithms, using single-waveband images or multi-waveband images, were developed to discriminate viable and nonviable seeds. The optimal wavebands for each algorithm to use were determined using correlation analysis. The single-waveband image (SWI) algorithm based on images at 684 nm yielded discrimination accuracies of 99.0% and 97.0% for viable seeds and nonviable seeds, respectively. The discrimination accuracies of the algorithms based on the subtraction image (S621-425), the ratio image (R-602/530), and the ratio-subtraction image (R520-698/602-698) were 100% and 99.0% for viable and non-viable seeds, respectively. The results demonstrate that hyperspectral fluorescence imaging has potential for quality sorting of cucumber seeds based on viability.