Determination of the sample number for optical reflectance measurement of vegetable leaf

다국어 초록 (Multilingual Abstract)

Visible and near-infrared spectroscopy provides a variety of information regarding leafy vegetables. However, various factors including optical propagation, environmental issues, and experimental issues affect the quality of spectral measurements. Therefore, stability and performance of optical reflectance sensors are affected significantly by sampling scheme. This study investigated the effects of the number of sampling on optical reflectance measurement of Chinese cabbage and kale plant leaves. For that purpose, variability and similarity of multiple measurements for different number of sampling of the leaves were evaluated. A combination of a median filter and the 2nd Savitzky-Golay (S. Golay) method was used to reduce the noise on the reflectance spectra introduced by background effects and other uncertainties. Reflectance difference and cross correlation were used as criteria to evaluate stability (or similarity) of the measurements. Results indicated that the standard deviation of reflectance difference was not considerably different for cabbage and kale leaves with 12 and 9 sampling points, respectively. For similarity of multiple measurements, results of cross correlation showed that the standard deviation of cross correlation was not greater than 5% with 12 and 9 sampling points for cabbage and kale leaves, respectively. Thus, this study concluded that 12 sampling points (cabbage) and 9 sampling points (kale) on a single leaf were the optimal number for spectral reflectance measurement. This study may provide guidelines on the number of sampling for optimum reflectance measurement of leafy vegetables.