2021 年 33 巻 3 号 p. 79-85
Near-infrared (NIR) spectroscopy has been used to predict fruit quality. However, thereʼs few previous studies which use small and lightweight devices and they didnʼt construct models for each cultivar although each tomato cultivar has its own characteristics such as color and size. In this study, models to nondestructively determine fruit quality of tomato cultivars A (orange-colored cherry tomato), B (red-colored cherry tomato), C (red-colored cocktail tomato), and D (orange-colored cocktail tomato) were constructed using an inexpensive, portable near-infrared (NIR) spectroscopic device (spectral response range: 590-1,050 nm). We irradiated the tomato fruits and analyzed the spectra of the scattered light. The spectral data were preprocessed using four different methods (min-max normalization, 1st differential, 2nd differential and combination of normalization and 1st differential). The soluble solid content (SSC) and citric acid content (CAC) of the fruits were determined by destructive methods. Partial least-squares regression was used to construct models for predicting the SSC and CAC from the preprocessed spectra. The best preprocessing method differed by cultivar. For the best and worst models of SCC, the R2 values were 0.80 and -0.1, respectively. Model performance may differ according to the color of the pericarp. Also, selection of an appropriate preprocessing method is important when using small spectroscopic devices to measure tomato fruit quality.
